Micro-CT scanners: Issues to do with calibration, accuracy, precision and standards. Why should we believe what they tell us?

Reconstructed images of a hand cross-section obtained with different preprocessing approaches (9) It seems that the trabecular structure of bones derived from fingers creates a texture.

a. Representative CT-image of a hand Result obtained from grey-scale mathematicl morphology image a) reconstruction (9)

A: Image of human femoral neck (There are many plates and struts lying orthogonal to them) (140)

G: Image displayed on Set up tab

Centre of Rotation panel.bmp

Figures - uploaded by Aikaterini Stamou

Content may be subject to copyright.

Content uploaded by Aikaterini Stamou

Content may be subject to copyright.

Aikaterini Stamou

Home institution: National Polytechnics School of Athens

Source: Osteoporosis: Bad to the Bone (Steven C. Mares, M.D. Williamsburg, Virginia)

Supervisor: Dr Peter Zioupos

This project was accomplished over CDS Summer Internship period lasting from

1st of June (of 2011) up to 31st of August (of 2011).

ACKNOWLEDGEMENTS

I wish to thank Doctor Peter Zioupos for making this research internship opportunity possible,

for his valuable support over this period and his help and enthusiasm throughout my project.

This enabled me to discover the field of micro-CT scanning of bones which is very interesting

and exciting.

I also acknowledge the help received by all the members of the Stevenson Laboratory. I

enjoyed the convivial atmosphere among the staff during my stay at Cranfield University –

Shrivenham.

Originality, liability and disclaimer statement

This is the text of a report by Aikaterini Stamou of the National Polytechnic

School of Athens, prepared after she completed a training internship in the

Dept of Engineering & Applied Science, in CDS (Cranfield Defence and

Security) of Cranfield University-UK and submitted to Cranfield University

in the Shrivenham campus describing her course of action during this

training period; the report is an internal document submitted to the two

aforementioned Institutions, the information of the text may contain errors

and omissions and thus the document does not have an official standing as

a ‘CDS-Cranfield’ or a ‘National Polytechnic School of Athens’ document;

all liability, originality and copyright for the experiments and data outlined in

this report stays with Cranfield University at Shrivenham and the academic

responsible for directing the biomechanical activities Dr. P. Zioupos.

P.Zioupos A. Stamou

Date: 07/09/2011

CONTENT OF THE STUDY

ACKNOWLEDGEMENTS ......................................................................................................... 1

CONTENT OF THE STUDY...................................................................................................... 3

INTRODUCTION ....................................................................................................................... 5

CHAPTER 1: STRUCTURE OF BONE TISSUE ...................................................................... 6

1.1. STRUCTURE OF BONE .............................................................................................. 6

1.2 COMPOSITION OF BONE.................................................................................................. 6

1.2.1 ORGANIC MATRIX .......................................................................................................... 7

1.2.2 MINERAL.......................................................................................................................... 7

1.2.3 WATER............................................................................................................................. 7

1.3 TYPES OF BONE................................................................................................................ 7

1.4 POROSITY OF BONE......................................................................................................... 8

1.5 STRUCTURAL PROPERTIES OF BONE........................................................................... 9

1.5.1 STRUCTURAL PROPERTIES OF CORTICAL BONE .................................................... 9

1.5.2 STRUCTURAL PROPERTIES OF CANCELLOUS BONE .............................................. 9

1.6 MECHANICAL PROPERTIES OF BONE ......................................................................... 10

1.6.1 MECHANICAL PROPERTIES OF CORTICAL BONE................................................... 10

1.6.2 MECHANICAL PROPERTIES OF CANCELLOUS BONE............................................. 10

CHAPTER 2 MEASUREMENTS IN BIOMECHANICAL LABORATORIES (CRANFIELD

UNIVERSITY, DEFENCE ACADEMY, UK) ............................................................................ 12

2.1 INTRODUCTION ............................................................................................................... 12

2.2 MEASURED SIZES WITH THEIR ASSESSMENT METHODS........................................ 13

In the recent study ................................................................................................................... 13

2.2.1 ASSESSMENT OF PHYSICAL PROPERTIES OF BONE ............................................ 13

2.2.2 ASSESSMENT OF MECHANICAL PROPERTIES OF BONE....................................... 14

2.2 RESULTS .......................................................................................................................... 14

CHAPTER 3: AN OVERALL REVIEW OF STUDIES USING MICRO-CT SCANNING

TECHNIQUES ......................................................................................................................... 16

3A. INTRODUCTION TO MICRO-CT TECHNOLOGY ........................................................... 16

DESCRIPTION OF TAKING MEASUREMENTS BY USING MICRO-CT TECHNOLOGY .... 16

3.A.1 THRESHOLDING TECHNIQUES.................................................................................. 16

COMPARISON BETWEEN GLOBAL AND LOCAL THRESHOLD TECHNIQUES................ 17

3B. PHANTOM FOR THE CALIBRATION OF MICRO-CT SCANNING IMAGES.................. 18

3.1 LITERATURE STUDIES EVALUATING RELATIONSHIPS BETWEEN BONE PHYSICAL

PROPERTIES WITH MICRO-CT TOMOGRAPHY................................................................. 19

PHYSICAL PARAMETERS (ARRANGEMENT OF) CHARACTERIZING BONE STRUSTURE

3.1.II ASH DENSITY VERSUS BV/TV BONE DENSITIES (APPARENT, TISSUE DENSITY)

......21

3.2.LITERATURE STUDIES EVALUATING RELATIONSHIPS BETWEEN BONE

MECHANICAL AND PHYSICAL PROPERTIES WITH MICRO-CT TOMOGRAPHY............. 23

CORRELATION OF MECHANICAL PROPERTIES TO PHYSICAL PROPERTIES OF BONE

STRUCTURE........................................................................................................................... 23

3.3.LITERATURE STUDIES EVALUATING THE ASSESSMENT OF MECHANICAL

PROPERTIES OF BONE BASED ON COMBINED USE OF COMPUTED TOMOGRAPHY

AND FINITE ELEMENT MODEL TECHNIQUES.................................................................... 24

3.3.I.SEGMENTATION AND CONTOUR EXTRACTION........................................................ 25

3.3.II ASSIGNMENT OF MECHANICAL PROPERTIES OF BONE USING FINITE ELEMENT

MODELS.................................................................................................................................. 26

GENERATION OF FINITE ELEMENT MESHES .................................................................... 26

ASSIGNMENT OF MATERIAL AND MECHANICAL PROPERTIES...................................... 26

3.4.FACTORS AFFECTING MEASUREMENTS OF MECHANICAL PROPERTIES BASED

ON THE MICRO-COMPUTED TOMOGRAPHY AND ON FINITE ELEMENT METHODS .... 26

LITERATURE REVIEW ........................................................................................................... 27

CHAPTER 4: MICRO-CT SCANNING ANALYSIS OF BONES ............................................ 29

SAMPLES USED FOR MICRO-CT SCANNING MEASUREMENTS AND SAMPLE

PREPARATION....................................................................................................................... 29

4.1 SAMPLE PREPARATION ................................................................................................. 29

4.2 DESCRIPTIONOF THE STUDY (Edmun Chun)............................................................... 29

4.3 MICRO-COMPUTED TOMOGRAPHY.............................................................................. 29

CHAPTER 5: PROCESS OF MICRO-CT SCANNING IMAGES............................................ 30

5.I CT-PRO SOFTWARE ........................................................................................................ 30

5.1 OVERVIEW OF USER INTERFACE................................................................................. 30

5.2 PROCESSING TABS OF USER INTERFACE.................................................................. 30

5.3 RECONSTRUCTION OF THE SAMPLE USING MICRO-CT SCANNING IMAGES........ 30

5.4 CALCULATION OF CENTER OF ROTATION.................................................................. 31

DEFINITION OF THE CENTER OF ROTATION .................................................................... 31

EXPLANATION OF THE FUNCTION OF CORRECTING CENTER OF ROTATION WITH CT-

PRO 32

USING THE CENTER OF ROTATION TAB (Figure 5.1.D) .................................................... 32

5.5 USING THE “SET UP” TAB............................................................................................... 33

5.6 USING THE “VOLUME” TAB ............................................................................................ 34

5.7 USING THE “CALIBRATION” TAB.................................................................................... 35

5.II VGLSTUDIOMAX SOFTWARE......................................................................................... 36

5.II.A PREPARATIONS (WARNINGS) DURING VGSTUDIOMAX USE ................................ 36

5.II.B WORKSPACE................................................................................................................ 36

OBJECT SELECTION FROM THE SCENE TREE................................................................. 37

5.II.C WORKSPACE WINDOW CONTROLS.......................................................................... 37

2-D WINDOWS........................................................................................................................ 38

5.II.D FILE MENU OPTIONS................................................................................................... 38

5.II.E OBJECT MENU OPTIONS ............................................................................................ 39

5.II.F SELECT MENU OPTIONS............................................................................................. 41

5.II.G ANALYSIS MENU OPTIONS ........................................................................................ 42

5.II.G CHARACTERIZATION OF THE QUALITY OF RECONSTRUCTED IMAGES ............ 43

6.GENERAL DISCUSSION..................................................................................................... 44

REFERENCES ........................................................................................................................ 45

A1. APPENDIX OF TABLES .................................................................................................. 54

Table 5.2.E: Analysis menu options A2. APPENDIX OF FIGURES................................... 65

A2. APPENDIX OF FIGURES................................................................................................. 66

A.2.I FIGURES DERIVED FROM FROM LITERATURE STUDIES ....................................... 66

A.2.II FIGURES DERIVED FROM OUR MICRO-CT SCANNING ANALYSIS....................... 82

IMAGES OBTAINED BY USING CT-PRO SOFTWARE........................................................ 82

IMAGES OBTAINED BY USING VGSTUDIOMAX SOFTWARE .......................................... 91

A3. APPENDIX OF EQUATIONS ........................................................................................... 95

INTRODUCTION

The mechanical properties (strength, Elastic modulus) of bones and their loading conditions

determine the risk of bone fractures.(118) In particularly, for elderly such bone fractures are

catastrophic evening , leading to an increased mortality rate and a decreased mobility(83, 14,

24).This situation is presented on elderly people due to the risk of osteoporosis on them but

also the greatly low rate of bone regeneration. An accurate and non-invasive assessment of

bone mechanical properties could be of high significance to diagnose the bone fracture risk

such that preventative measures can be taken at time. Current methods are based on bone

densitometric measurements. (118)

The mechanical properties of bones are determined by the external and internal architecture

of bone and by bone tissue properties.(118) The assessment of mechanical properties of

trabecular bone regions is of particular significance because the osteoporotic fractures initially

appear in these regions. Bone densitometric measurements can provide information about

the bone mineral density of these regions which is related to bone volume fraction. (118) It

needs to be stated that mechanical properties of this type of bone tissue are not determined

by its bone volume fraction alone(118) but also by parameters related to the arrangement of the

structure of trabecular bone material or its architecture. Thus, a better prediction of bone

mechanical properties is achieved by taking into account these architectural factors.

Among the non-destructive techniques of evaluating architecture of bone, X-ray micro-

computational tomography (micro-CT) has been widely used for the characterization of the

trabecular bone structure. (6) This is beneficial for the assessment of trabecular bone(104) tissue

due to its being a non-destructive method, avoids specimen preparation and provides three-

dimensional (3-D) images with a high and isotropic spatial resolution up to a few micro-meters

in the three spatial directions(153).

Calculations based on micro-Finite Element (micro-FE) techniques applied on scanning

images (images of bone specimens obtained by using micro-CT scanner).(118) It could be said

on another way that this application of micro-FE techniques is based on high-resolution 3-D

images of bone specimens, captured using micro-CT or other high-resolution techniques(34,20)

and reconstructed in a computer. These reconstructions are simulated by using micro-FE

models and thus represent the internal architecture of bone in detail. (118) By solving the FE-

problem for different loading patterns the mechanical properties of bone specimens can be

estimated on the whole bone.

The purpose of the study described through this report is using of micro-CT scanning

technique for the evaluations of parameters characterizing the trabecular bone micro-

architecture. In specific, we would like to verify relationships between structural properties of

trabecular bone specimens which have been included in publicized studies referred to using

of CT-techniques. It would be done by corroborating reconstructed X-ray micro-CT images

through the use of software packages processing CT-imaging.

CHAPTER 1: STRUCTURE OF BONE TISSUE

The human skeleton carries out a number of essential roles which enable the human body to

function. Firstly, It is mainly important to be said that It plays a vital part in the musculoskeletal

system by providing insertion points for muscles and tendons which allows for locomotion and

other muscle action. Furthermore, human bone tissue acts as a protective barrier for the vital

organs such as the heart and brain while it provides a network to support them. In addition, it

acts as a storage supply for crucial minerals contributing in the existence of life such as

calcium, magnesium and phosphorus. Lastly, the medullary canals being presented in the

interior of the bone provide the site for the production of the body’s blood cells from the bone

marrow.

1.1. STRUCTURE OF BONE

It would be regarded as main concern for each researcher to apprehend the mechanical

properties of each component phase of the bone as well as the structural relationship

between them at the various levels of hierarchical structural organization (in order to

apprehend the mechanical properties of the bone material (34). These levels and the

respective structures are the following:

(1) Macrostructure: cancellous and cortical bone

(2) Microstructure (10-500μm): Harvesian systems, osteons, single trabeculae

(3) Sub-microstructure (1-10μm): lamellae,

(4) Nanostructure (from a few hundred nanometers to 1μm): fibrillar collagen and embedded

mineral

(5) Sub-nanostructure (below a few hundred nanometers): molecular structure of constituent

elements (e.t.c. mineral, collagen and non-collagenous organic proteins)

This hierarchically organized structure has an irregular, yet optimized, arrangement and

orientation of the components, making the material of bone heterogenous and anisotropic

(Figure 1.1)(112).

Figure 1.1: Hierarchical structural organization of bone(6)

1.2 COMPOSITION OF BONE

The compositional makeup (organisation, microstrustructure) of bone can be partitioned into

three distinct components, organic part (mainly constisted of collagen), mineral (or inorganic

part) and water (≈10% of the bone mass) (9,20), with the percentage of each varying depending

on the species and the requirements of the bone to fulfil its job(23). A tertiary diagram showing

the range of variation that occurs naturally between species, with the mineral content ranging

from as little as 39.3% (Red deer antler) to 96% (Mesoplodon rostrum) is figured by Figure

1.2(140).

1.2.1 ORGANIC MATRIX

Collagen is a protein which can organize itself into fibers. The predominant form of collagen

found in mature bone is of Type I. This component gives the bone matrix its flexibility and

tensile strength and provides area for mineral crystals. (113) The existence of non-colagenous

proteins (≈10% of the organic matrix) in the organic matrix has a high impact on bone-cell

function rather than on the mechanical properties of the bone (1, 45, 99,113).

The non-collagenous proteins are composed of proteoglycans, osteocalcin, phospholipids,

osteopontin and the bone morphogenetic protein. Concerning its function, the proteoglycans

and osteocalcin are linked with the remodelling process, (13, 8) the phospholipids and

osteopontin are linked with the degree and the control of mineralization within the tissue,(8)

while the bone morphogenetic protein is linked to osteoinductive properties.(8)

1.2.2 MINERAL

The collagen network forms the scaffold for the deposition of the mineral

matrix which fills the 40nm gaps between the collagen molecules, and packs the spaces

between the collagen fibrils.(23) The mineral is a mixture of calcium phosphate (Ca3(PO4)2),

calcium hydroxide (Ca(OH)2) and phosphate ions in the form of crystals of hydroxyapatite

(Ca10(PO4)6(OH)2). The hydroxyapatite is, impure and incorporated into the crystals which are

other ions and compounds such as HPO4, Na, Mg, citrate, carbonate and K.(112)

1.2.3 WATER

The water content of the bone matrix is partially bonded to collagen. However, free water is

presented as well and takes part in the mineralization process.(113)

1.3 TYPES OF BONE

Independent of their macroscopic anatomy, all skeletal segments consist of an outer layer of

compact bone (called as cortical bone) and an inner zone (the medulla) that contains bone

marrow. The relative proportion among the compacta and the medulla varies with the skeletal

segments and their function. The diaphysis of long bones displays a thick compacta, which

about 90% of the volume is calcified in and a medulla which corresponds to an axial, more or

less eccentric cylinder containing red, hemopoietic bone marrow in youth and yellow, fat

repleted, nonhemopoietic marrow in adults. The medulla consists of a framework of

interlacing laminar or osseous trabeculae.(8)

The outer compacta of the skeletal segments consists of compact (cortical) bone (Figure

1.3A); the inner medulla corresponds to the bone marrow cylinder in long bones, to interlacing

osseous trabeculae in short and flat bones and in long bone epiphyses. The osseous

trabeculae form the cancellous or trabecular bone. Comparing the two types of bones

according to their porosity, it could be stated that cortical bone is much denser with a porosity

ranging between 5% and 10 %.(152)

The cortical bone is arranged in a hierarchical structure being composed mainly of the

following levels of structure (Table 1.3A taken from Appendix):

A. First Level Cortical Bone Structure(152)

There are four types of different structural organizations, being described as the 1st structural

level which are called woven bone, primary bone, plexiform bone, and secondary bone.

A.1 Woven-fibered cortical bone(152)

Woven bone does not contain osteons as does primary and secondary bone, nor does it

contains the brick-like structure of plexiform bone. Thus, woven bone is the most disorganized

of bone tissue owing to the circumstances in which it is formed. Woven bone tissue is the only

type of bone tissue which does not need to form on existing bone or cartilage tissue.

A.2 Plexiform Cortical Bone Tissue(152)

Like woven bone, plexiform bone is formed more rapidly than primary or secondary lamellar

bone tissue. However, unlike woven bone, plexiform bone should offer increased mechanical

support for longer periods of time

A.3 Primary Osteonal Cortical Bone Tissue(152)

When bone tissue contains blood vessels surrounded by concentric rings of bone tissue it is

called osteonal bone. The structure including the central blood vessel and surrounding

concentric bone tissue is called an osteon. Primary osteons are likely formed by

mineralization of cartilage, thus being formed where bone was not present. As such, they do

not contain as many lamellae as secondary osteons.

A.4 Secondary Osteonal Cortical Bone Tissue(152)

Secondary osteons differ from primary osteons in that secondary osteons are formed by

replacement of existing bone. Secondary bone results from a process known as remodeling.

In remodeling, bone cells known as osteoclasts first resorb a section of bone in a tunnel

called a cutting cone. Following the osteoclasts are bone cells known as osteoblasts which

then form bone to fill up the tunnel. The osteoblasts fill up the tunnel in staggered amounts

creating lamellae which exist at the 2nd level of structure.

B. Second Level Cortical Bone Structure(152)

The second level cortical bone structure consists of those entities which compose the osteons

in primary and secondary bone and the "bricks" in plexiform bone. Within osteonal (primary

and secondary) and plexiform bone the four major matrix 2nd level structural entities are

lamellae, osteocyte lacunae, osteocyte canaliculi, and cement lines. The lamellae are

arranged concentrically around the central haversian canal in osteonal bone. Cement lines

are only located in secondary bone because they are the result of a remodeling process by

which osteoclasts first resorb bone followed by osteoblasts forming bone. The cement line

occurs at the point bone resorption ends and bone formation begins.

Trabecular bone being the second type of bone tissue in the body, fills the end of long bones

and also makes up the majority of vertebral bodies. According to cortical bone, trabecular

bone structure is organized having a criterion physical scale size. (152)

The major divergence between trabecular and cortical bone structure is found on the 1st and

2nd structural levels (Table 1.3A, Table 1.3B taken from Apendix). Trabecular bone is more

compliant than cortical bone and it is believed to distribute and dissipate the energy from

articular contact loads. Trabecular bone contributes about 20% of the whole skeletal mass

within the body while cortical bone contributes the remaining 80%. (152)

C. COMPARISON BETWEEN CORTICAL AND TRABECULAR BONE (152)

One of the largest differences between trabecular and cortical bone is noticeable at the 1st

level structure (Table 1.3A, Table 1.3B taken from Apendix). As being displayed in the first

table, trabecular bone is much more porous than cortical bone. The basic structural entity at

the first level of trabecular bone is the trabecula. Trabecula are most often characterized as

rod or plate like structures. Unlike osteons, the basic structural unit of cortical bone,

trabeculae in general do not have a central canal with a blood vessel.

The 2nd level structure of trabecular bone has most of the same entities as the 2nd level

structure of cortical bone including lamellae, lacunae, canaliculi, and cement lines. (Table

1.3A, Table 1.3B taken from Apendix) Trabecular bone does not generally contain vascular

channels like cortical bone. Trabecular bone is differentiated from cortical bone structure

concerning the arrangement and size of these entities. For instance, although lamellae within

trabecular bone structure are of approximately the same thickness as cortical bone (≈3

μm(69)), the arrangement of lamellae is different. Lamellae are not arranged concentrically in

trabecular bone as in cortical bone, but are rather arranged longitudinally along the trabeculae

within trabecular packets(69).

1.4 POROSITY OF BONE

Two of the most important features of the internal structure of bone are porosity and specific

surface (being discussed in detail in Chapter 3).(0) Porosity is defined as void volume per unit

volume of whole bone or the fractional part of bone occupied by soft tissues.

In first, a general description of he geometry of the porosity of the bone will be given.(19) A

sketch of cortical bone is shown in Figure 1.4.

The major voids in this type of bone are canals which twist and wind through the specimen

with considerable irregularity.(27) In general, they go around and into the bone.(13) They

bifurcate and anastomize frequently and at various angles, but they have been divided into

two groups, Haversian canals which run nearly parallel to the long axis of the bone and

Volkmann’s canals which run perpendicular to the long axis.(0) In addition to these major

voids, the bone is filled with many small, ellipsoidal holes, named lacunae containing cells

named osteocytes.(1) These voids are connected to other and the larger canals by extremely

fine tunnels (≤0.5μm in diameter), known as canaliculae.(45)

In trabecular bone, the voids dominate the space. (1) The soilid matrix of the trabecular bone is

reduced to a complex system of interrupted walls and struts called trabeculae. The voids are

called marrow spaces due to their major contents. Inside the trabeculae, voids could be found

being so small as those ones which could be observed in cortical bone, lacunae and

canaliculae. In both cortical and trabecular bone almost all the voids are connected because

they are essentially a network of vascular spaces.(1)

1.5 STRUCTURAL PROPERTIES OF BONE

1.5.1 STRUCTURAL PROPERTIES OF CORTICAL BONE

The mechanical properties of cortical bone are affected greatly by the porosity, the

mineralization level and the organization of the solid matrix. The bone density, bone mineral

content, surface area and porosity are considered as the main structural properties of the

bone and they are defined on the following text.

1.5.1.A BONE DENSITY AND BONE MINERAL CONTENT

The material density of cortical bone is the wet weight divided by the specimen volume. It is a

function of both the porosity and the mineralization of the minerals of the bone. For cortical

bone, apparent and material density are basically the same due to the absence of marrow

space in compact bone (cortical bone). Thus, “cortical bone density’ is used to describe the

density of cortical bone.(1) A positive correlation is existed between apparent density of cortical

bone and its mechanical properties. (1)

The true meaning of bone mineral density (BMD) is bone mineral mass per unit volume.(1)

(Equation 1.5.1)

Bone

eral

BMD min

= (Equation 1.5.1)

where: eral

mmin is bone mineral mass

Bone

V is bone volume

Similarly, the term bone mineral content (BMC) is described as the ratio of unit weight of the

mineral portion to dry bone unit weight and is frequently expressed as a percentage. (1)

(Equation 1.5.2)

1.5.1.B SURFACE AREA

Surface area (Specific surface) of cortical bone is defined as the internal surface area per unit

volume of whole bone. (1)

1.5.1.C POROSITY

It could be easily understood that a more porous bone has a weaker mechanical strength.

Porosity is defined as the ratio of void volume to total bone volume.(1)

1.5.2 STRUCTURAL PROPERTIES OF CANCELLOUS BONE

The porous matter of cancellous bone being surrounded by trabecular columns and marrow-

filled pores or cavities can be described by using a set of structural properties.(1) It could be

considered as main structural properties the bone density, the trabecular bone volume

fraction, the porosity, the trabecular bone area, the trabecular thickness, the trabecular

separation and the trabecular number.

1.5.2.A BONE DENSITY AND BONE MINERAL CONTENT

A strong correlation exists between the mechanical properties of cancellous bone, both for

strength and stiffness.(Table 1.5a, Table 1.5b). According to (Sharp, 1990)(126) the two

densities which are defined for trabecular bone are termed.

Apparent density is the mineralized tissue mass per total volume of the sample and is a

function of the amount of bone present. (1) Real density(1) is the mineralized tissue mass

divided by the volume of the matrix excluding the marrow vascular spaces and is thus the

density of the bone matrix by itself.(1)

1.5.2.B BONE VOLUME FRACTION

Bone volume fraction (Relative bone volume) is defined as the proportion of a studied volume

of trabecular bone occupied by bone as opposed to marrow and is expressed by

BV (Bone volume/Tissue volume).(155)

It is assumed that trabecular bone has bone volume fraction ranging from just over 5% to a

maximum of 60%.(154) The trabecular bone volume fraction varies between different bones,

with age, and between species.(154)

1.5.2.C POROSITY

Porosity (

) can be expressed as one minus the ratio of apparent density to real

density(124).((Equation 1.5.3)

1.5.2.D TRABECULAR BONE AREA

Trabecular bone area is defined as the trabecular surface area divided by the total area

(expressed in μm2). (155)

1.5.2.E TRABECULAR THICKNESS

Trabecular thickness ( ThTb.) is defined as the average thickness of tracebulae (measured in

μm).(1) The estimation of ThTb.from 2D (two-dimensional) measurements requires an

assumption about the nature of the bone structure.

1.5.2.F TRABECULAR SEPARATION

Trabecular separation ( SpTb.) is substantially the thickness of the marrow spaces between

trabecular structures.(155)

1.5.2.G TRABECULAR NUMBER

It could be considered that Trabecular number ( NTb.) implies the number of traversals

across a trabecular structure made per unit length on a linear path through a trabecular bone

region.(155)

1.6 MECHANICAL PROPERTIES OF THE BONE

The mechanical properties of bone depend on bone microstructure among other factors(1).

The main mechanical properties of the two types of bone tissue are elastic modulus (Young’s

modulus), ultimate stress, yield stress and Poisson’s ratio.

1.6.A ELASTIC MODULUS

Elastic modulus (or Young’s modulus (

) is determined from the slope of the initial linear

part of the stress-strain curve describing the mechanical behavior of the bone.

1.6.B ULTIMATE STRESS

Ultimate stress (ultimate strength) is defined as the maximum stress value that the bone can

withstand while being stretched without being subjected to fracture.

1.6.C YIELD STRESS

Yield stress is the equivalent stress value below which the bone behaves like a solid material.

1.6.D POISSON’S RATIO

Poisson’s ratio ( v) is defined as the ratio of the strain being caused in the transverse

direction per unit of the strain being caused in the axial direction o the bone structure.

1.6.1 MECHANICAL PROPERTIES OF CORTICAL BONE

The mechanical properties of bone are affected greatly by the porosity, the mineralization

level and the organization of the solid matrix.(112) Values of mechanical properties of bone at

the macrostructural level vary from one bone to another as well as within different regions of

the same bone.(114,34) For a number of human bones (including the femur, tibia, humerus,

mandible, lumbar vertebrae and patella) the site-specific orthotropic elastic modulus, shear

modulus, Poisson’s ratio have been studied as a function of position(114,115) using ultrasonic

techniques. The mechanical properties of human cortical bone from the tibia, femur and

humerus have been found to vary between subjects, even if the density is the same.(112)

Concerning the phenomenon of anisotropy, the elastic modulus in the longitudinal direction

were not very different between the various types of cortical bone and a greater modulus

variability existed along the length of a whole bone than around its circumference.(112) The

elastic modulus in radial or circumferential directions correlated to the longitudinal modulus,

but the correlation coefficients were relatively low (of the order of (0.001-0.06).(112)

1.6.2 MECHANICAL PROPERTIES OF CANCELLOUS BONE

In human cancellous bone, there is no difference in the mechanical properties of the

humerus, the proximal tibia and the lumbar spine.(114,115) The stiffness and strength of

cancellous bone from these bones were found to be lower than those of the patella, and the

distal and the proximal femur (patellar cancellous bone had the highest values overall. In

human cancellous bone, mechanical properties differ significantly around the periphery and

along the length and significant inter-subject differences are shown.(112) The differences in

inter-bone mechanical properties suggest that predictions of mechanical properties can not be

explained by taking into consideration only the density of cancellous bone.(112)

CHAPTER 2 MEASUREMENTS IN BIOMECHANICAL PROPERTIES

The content of this chapter is focalized on a review of the experimental measurements by

using samples of femoral bone which took place in Biomechanical laboratories (Cranfield

University, Defence Academy , UK). In specific, a brief introduction is given initially based on

a theoretical background and previous studies intending to facilitate the comprehension of the

results of measurements which were obtained in these laboratories. In addition, the methods

of measurement of physical properties of the bone are described and a display of the results

follows.

2.1 INTRODUCTION

Density is a notice worthy property of bone and plays and a crucial role in determining the

mechanical properties of cortical and cancellous bone tissues.(140)

Density is defined in a number of ways at either the bone tissue ( app

Dapparent) or the bone

material level ( mat

D material). The “apparent density” value is defined as the wet mineralized

mass of bone at the tissue level divided by the volume occupied by the tissue. The “real” or

“material” density of bone material is termed as the wet mineralized mass of tissue divided by

the volume occupied by the material itself. The difference among them exists due to the

presenc of porous or vacuous spaces which are related to the canaliculi, osteocyte lacunae,

osteonal canals and similar non-mineralized architectural features.(140)

Measuring the density in the context of the bone is a complex issue. There is a partial

argument among studies on the matter of the particular attention that must be paid to the

flushing of the liquids from the pores and the subsequent re-filling of them with the

suspending fluid in case of applying Archimede’s principle.(2) It is suggested by other studies

that defatting and/or preserving the marrow may have and effect if it interferes with the

re(de)hydration process.(124) Substantial differences have been referred by subsequent

studies being presented in density values obtained by the invasive/in-vitro Archimedes’

principle and the non-invasive DXA (Dual X-ray Absortptionmetry) or micro-CT (micro-

Computing Tomography) methods.(61)

It is supported by the study.(140) that the concept of porosity

(

)

P−= 1with TV

BV as

the Bone volume fraction, provides a geometric and visible measure of the level of porosity in

the structure. It could be stated that the value of TV

BV is relative to the value of app

D and

mat

D. Two obviously different cancellous bone structures can have the same nominal

BV for properly altered apparent density vs. material density values. (140)

app

D has a primary impact on the mechanical properties at the tissue level(117), while mat

determines material behavior at the trabecular level and by implication properties at the

structural level. TV

BV features prominently in experimental studies of cancellous bone

mechanics, but also indirectly in theoretical studies as it relates linearly to the ratio of

mat

app D

D(models described in model suggested by (Gibson and Ashby, 1997)(34)).

Latest derived data from Biomechanicall Laboratories (Cranfield University, Defence

Academy , UK) revealed an interdependence of app

Dto mat

Din at least four different cohorts

of patients who underwent surgery followind either osteoporotic or osteoarthritis

complications.(152,150,19)

Plots of app

D versus mat

D show strong negative correlations.(Figure 2.1)(140) As it is

displayed on this, increases in apparent density are accompanied with associated decreases

in material density of the trabeculae. This is in itself constitutes a paradox as assumingly for a

common bone matrix density of 2.2 g cm-3, the end point of the process of gradually reducing

pore sizes and thus increasing apparent density should also be 2.2 g cm-3.(Figure 2.1).

A small number of data points from OA material showed such a trend, but it is uncertain

whether it is an effect of abnormalities observed in OA matrix which shows proliferation of

osteid tissue and compact like bone areas.(140)

It was impossible being found any literature studies referred similarly to an inverse app

D-

mat

D relationship but at least three independent studies were identified where the progress of

material trabecular density showed an inverse effect to the structural mechanical integrity of

cancellous bone (38,97,88), thus corroborating original observations derived from Biomechanical

Laboratories (Cranfield University, Defence Academy , UK). Due to existing uncertainties

about the causes of the displayed behavior(Figure 2.1) it was assumed that the effects are

either specific to the structural form of bone (cortical vs. cancellous) or specific to the

condition associated with the nature of tissue (OP/OA/healthy donors(19)) or they may be due

to inter-individual vs. intra-individual effects.(150) Such queries can only be elucidated by

producing equivalent relationships from specimens originating from one and the same healthy

bone to remove these confounding factors.(140)

In the below text, the methods of measurement of physical properties of the bone and the

results of a recent study(140) which inspected the basic relationships between density values in

cortical and cancellous bone regions derived from the femur of Asian elephant, are described.

2.2 MEASURED SIZES WITH THEIR ASSESSMENT METHODS

In the recent study(140) mechanical property (Young’s modulus (

)) and physical properties

including apparent density ( app

D), trabecular material density ( mat

D), Porosity (

) and Bone

volume fraction

(

)

BV of cancellous and cortical bone of the Asian Elephant femur were

assessed by a set o methods as it is followingly described. In this study the assessment of

V, eight

W, sub

Wis essential for the subsequent assessment of app

D, mat

and

(

)

BV .

2.2.1 ASSESSMENT OF PHYSICAL PROPERTIES OF BONE

Each sample (cortical or cancellous bone tissue) was cleaned using a high-pressure water jet

to remove any bone marrow and fat and then left for 48hours in a solution of 1; 1 chloroform/

ethanol to dissolve any remaining fat. After 48hours, the cubes were rehydrated gradually and

washed with water/ethanol mixtures. Then, they were left to rehydrate fully for a further 24h in

Ringers solution.(140)

The dimensions of each specimen were measured using a Vernier calipers in order to

produce a volume assessment

()

Weights were measured by use of an electronic microbalance either in air

(

)

eight

W or in

submersion

()

sub

Wusing a liquid of known density (distilled water, density ≈1g/cm3).

In practice the samples were first degassed thoroughly, then their submerged weight was

measured and then they were weighed in air. Before the

(

)

eight

W was taken, samples were

placed in a centrifuge for 3-min with a speed of 3000revisions/s to remove excess amounts of

water from their major pores. (140)

The measures of apparent density, trabecular material density, porosity and bone volume

fraction

(

)

BV are assessed by using following equations one per one in each case

Apparent density o

weight

app V

D=(Equation 2.2.1.1)

Trabecular material density

()

subweight

weight

mat WW

D−

(Equation 2.2.1.2)

Porosity (%) ⎥

⎦

⎤

⎢

⎣

⎡⎟

⎠

⎞

⎜

⎝

⎛

−= mat

app D

P1100 (Equation 2.2.3)

Bone volume fraction mat

app D

TVBV =/(Equation 2.2.1.4)

After mechanical testing a smaller size sample in comparison with the used ones for

mechanical testing, was removed from each sample, weighed in its hydrated state and

dehydrated test, then ashed at 600oC for 12hours before being weighed for a third time, Then,

the determination of the water

()

%Wat , mineral

(

)

%Min and organic

(

)

%Org content

fractions of each individual sample over its initial wet weight.

2.2.2 ASSESSMENT OF MECHANICAL PROPERTIES OF BONE

Before taking the mechanical measurements, the core samples were held in grips with a

miniature contact extensometer attached to a them at a central 6mm gauge. They were

constantly irrigated with Ringer’s solution at 37oC, were preconditioned for a few cycles to a

max strain of 0.1% and eventually, they were taken to just beyond the macroscopic yield so

that a definition of Young’s modulus, ultimate stress and yield stress is obtained.(140)

Additional 112 non-destructive tests were performed on the cube-formed specimens between

two polished loading platens contained within a Ringer’s bath system set up to maintain

physiological conditions throughout testing.

These tests allowed measurements of mechanical properties along three orthogonal axes (by

observing the degree of anisotropy) and used an external extensometer. (140)

2.2 RESULTS

Figure 2.2.1 shows the behaviour of Dapp vs. Dmat for all samples collected from the one

elephant femur. It is shown by the data a formed “boomerang”-like pattern with an inflection

point at about Dapp≈1.3 gcm-3 and Dmat≈1.6 gcm-3. The horizontal drawn line was an eyeball

estimate to separate the data into to regimes. Samples of Dapp>1.3 gcm-3 show a positive

correlation between the densities; those below show an inverse relationship. It was shown by

using qualitative inspection of the samples above and below the threshold that they were

derived from cortical area (or areas consisting of compact bone) and cancellous areas

respectively. Starting from a point of minimum porosity (inntracortical value) Dmat seems to

reduce to a value of Dmat =1.5 gcm-3 and then reverses order towards higher values for the

most porous structures. (140) The latter comes in agreement with previous studies(152,150) which

were exclusively on human studies.

Figure 2.2.2 displays material density of bone being explained by a rise in the mineral

content. The two datasets for trabecular and cortical bone can be described by one

relationship; they overlap remarkably over the full range of Dmat. It could be said on another

way that it is showh through this figure that mineral content has a single linear relationship to

the material density (Dmat) for both cancellous and cortical bone areas with a significant

overlap throughout the range.(140)

Figure 2.2.3 shows elastic modulus

(

)

Evalues produced vs Dapp, Dmat and BV/TV. It is worthy

being point out that

is increased as a high power vs all these three variables. In the case of

Dmat (Figure 2.2.3.b) there is a significant overlap between values for compact and

cancellous bone areas. It could be stated that there is a greater scatter and whether exact

elastic modulus

()

Evalues are demanded they would depend on whether the volume was in

fact cortical or cancellous bone. For Dapp<1.3 gcm-3, the

(

)

Evalue is practically indeterminant

by Dmat alone and has more to do with the structural effects in cancellous bone (Figure

2.2.3.a, Figure 2.2.3.b, Figure 2.2.3.c). For Dapp>1.3 gcm-3,

is a high power of Dmat.

(Figure 2.2.3.a, Figure 2.2.3.c) display an encouraging result for Dapp and BV/TV because

there is no overlap between

and values for cortical and cancellous bone. (140)

Figure 2.2.4 shows up a bimodal relation between Mineral content

(

)

%Min and

(

)

BV due to the tight relationship between Dmat and

(

)

%Min .(140)It could be claimed that

the lines figured “denote” an envelope being drawn by hand.

CHAPTER 3: AN OVERALL REVIEW OF STUDIES USING MICRO-CT SCANNING

TECHNIQUES

3A. INTRODUCTION TO MICRO-CT TECHNOLOGY

Microcomputed tomography has been widely used to analyze 3-D structure of bone(140) and

specific characterize trabecular bone structure. (6,140) Trabecular bone which exists in several

human bones, appears in as an interconnected network of rods and plates. Morphometrical

analysis of trabicular bone has been used in early diagnosis and assessment of osteoporosis.

The morph metrical, topological and geometric properties of bone networks are likely to play

important roles in determining bone strength. (104)

DESCRIPTION OF TAKING MEASUREMENTS BY USING MICRO-CT TECHNOLOGY

Estimation of morphometrical indices requires an accurate segmentation of trabecular

material in CT imaging as each greyscale level corresponds to a specific material density.(140)

Thus, threshold techniques have been broadly used in segmentation of bone images.(75, 101,

120, 45, 26) The key issue of these techniques is selecting a suitable threshold for separating

bone from soft tissues and background(6) and thus they are widely called as segmentation

techniques. The segmentation techniques are implemented on the reconstructed images of

the bone in which the grey value of each voxel represents an attenuation co-efficient in

Hounsefield units, into binary mages that only represent bone and non-bone. In other words,

for each voxel in the dataset a desicion needs to be made iwhether a voxel is bone or not.(19)

A set of different threshold techniques has been described in number of publisicized

studies.(26, 138, 32, 33, 69)

The setting of scales in a reconstructed image is considered as definitely necessary for the

assessment of bone density and parameters characterizing the bone microstructure and their

determination as a function of scanning parameters and particularly Hounsefield units (HU).

This can be accomplished efficiently by using the calibration procedure of the bone samples.

Calibration of bone samples requires on precedent stage the scanning of a cylinder including

material of the same bone density (widely named as phantom) as our bone samples. This

physical 3-D phantom needs being characterized of different known geometries and

thicknesses resembling those of the examined structures.(104) A number of studies(104, 56, 125)

has been focalized on search for a phantom used for an efficient calibration of the bone

specimens.

In this short chapter, an overview of threshold techniques and physical phantom used for the

calibration of X-ray micro-CT reconstructed images is done. This overview of threshold

techniques and phantoms is based on a number of publicised studies.

3.A.1 THRESHOLDING TECHNIQUES

An important theme for micro-CT analysis of bone samples is the segmentation of the original

greyscale reconstructed images being produced by mcro-CT scanning in order to separate

bone from non-bone.(138) The existence of problems such as noise, resolution limitations and

beam-hardening make this a non-trivial issue. Inappropriate segmentation methods may

reduce the potential power of micro-CT and introduce bias in the assessment of bone

structure parameters , specifically in case of in vivo scanning, where scanning time should be

as short as possible so long provided that a limited radiation dose is absorbed by the living

tissue.

The most widely used segmentation methods use global thresholds. Then, a single CT- unit,

which is commonly the Hounsefield unit is chosen, above which all voxels (3-D pixels) are

indicated as bone and below which all remaining voxels are idicated as non-bone. The used

value as a threshold is selected either visually by analyzing the histogram of Hounsefield

Units (HUs) or by forcing the resulting binary data to have the same volume as the original

bone sample measured by Archimede’s test(26). Concerning the entire bones, it is even

practically impossible to assess the real volume of the bone counted on Archimede’s

principle. This state excludes the global segmentation method based on knowledge about the

bone volume.

Despite the easy using of a global threshold, serious problems such as beam hardening,

noise and partial volume effects can considerably reduce the quality of the segmentation.

These factors can affect significantly the quality of the reconstructed images and in addition,

the calculation of architectural parameters describing the bone structure. The combination of

these effects (e.t.c. beam hardening, partial volume effects) bring about the optimal threshold

value for a certain part of the reconstruction differing from the optimal value in other parts. It

could be generally said that using a single global threshold value will result in the loss of thin

trabeculae and oversizing of thicker trabeculae.

Segmentation can be improved by using local threshold values than a single global threshold

so that each voxel can be thresholded optimally within its neighbourhood. Concerning the

literature studies searching for the local threshold techniques, Dufesne(32) has developed a

local threshold algorithm for CT-scanned imaging to compensate for beam-hardening effects

based on the analysis of the histogram of the local neighbourhood of a voxel. According to

this study, the efficiency of the local thresholding method possibly decreases when the

resolution is restricted and tissues are heterogenous. Kuhn et al.(69) and Elmoutaouakkil(33)

proposed more general local segmentation methods, both based on the “one-half maximum

height (HMH) protocol”. In brief, it could be said claimed than in “one-half maximum height

(HMH) protocol”, voxels are considered bone if their HU-number is higher than one-half the

difference beteen local minima (background) and local maxima (bone).

In following passage, a comparison between global and local threshold segmentation

techniques is described.

COMPARISON BETWEEN GLOBAL AND LOCAL THRESHOLD TECHNIQUES

On the below text, using of global threshold techniques is compared the local threshold ones

for the segmentation of micro-CT reconstructed images. At first, the global segmentation

method can give a too nice segmentation of greyscale reconstructed images where thick and

thin trabeculae are presented with the correct thickness and the structural integrity of

trabecular network is represented correctly without interruptions of the connected trabeculae

under prior knowledge of the real volume of the bone.(138) This is a not needed perquisite for

the automatic local threshold algorithm. When scanning of core biopsy specimens becomes,

the bone volume can only be obtained as a result from a cumbersome measurement

procedure based on Archimede’s principle. However, for the entire bones, it is practically

impossible to get the real volume of the bone and thus, using the global segmentation method

based on knowledge about bone volume is deterred. (138) It could be impossible to obtain the

real volume of a bone when animals are scanned in vivo. Because the radiation load on living

tissues should be as small as possible, the quality of scans is limited(138). As an effect of this

state, the resulting images contain more noise and have a lower resolution than those

obtained from conventional micro-CT scans. This situation places high requests on the

method of thresholding. Although the influence of noise seems rather small on both local and

global segmentation methods, the second ones fail to give a good representation of the bone.

This is partly created by the relative low resolution of the system that makes structures

appear less dense. (138)Besides, the result of global thresholding would be affected by

differences in mineralization between cortical and trabecular bone.(Figure 3A)

When good-quality scans are made at high resolution and the samples are of homogenous

structure, a global threshold performs just as well as local the local threshold method. A

typical situation as its being described is the case of the existence of bone biopsy specimens

being scanned at high resolution. (138) The convenience and speed of applying a global

threshold makes using this method very tempting. However, the implications of selecting the

global threshold should not be underestimated. In most studies, bone biopsy specimens of

subjects with a certain pathological condition or biopsy specimens resulting from some

intervention study are compared with controls. The possible changes in bone bone

morphology and bone mineralization caused by these pathologies or interventions would

influence the distribution of densities of the scans and thus they would probably interact with

the selection of threshold value. In general, using a global threshold might result in

uncertainties about which part of the assessed difference between groups are caused by the

selection of a threshold value. (138) Because the result of the local threshold methods is not

inspired by changes in mineralization and it is less sensitive to changes in architecture

(mainly to changes in the amount of thick versus thin trabeculae) using this method could

reduce the uncertainty about measured divergences among groups.(138)

According to literature,(138) a decrease in resolution to a voxel size of 53μm resulted in

unreliable results for global and local segmentation techniques. It was shown on this study(138)

that a local segmentation method (“LocalAuto method”) (138) gave better volume estimates

than a global segmentation method (“Global Hist method”). A study by Laib and

Ruegsegger(0) attempted to extract volume-related parameters from human in vivo scans at

much lower resolutions than the study(138) (165μm voxel size). Their methods gave

surprisingly fine reliability values for the measured parameters, comparable with the global

methods(138). Nevertheless, the low resolution made it impossible to extract the exact

structure.

3B. PHANTOM FOR THE CALIBRATION OF MICRO-CT SCANNING IMAGES

X-ray micro- computed tomography has rapidly gained importance as a non-destructive

investigation technique, especially in the 3-D examination of trabecular bone structure. By

intending to evaluate the accuracy of three-dimensional parameters determining the

trabecular bone structure, a three-dimensional phantom of well known material (known

density), shape and linear dimensions resembling those of the examined bone structures is

(would be) needed. This phantom is scanned before the scanning of the bone samples. This

procedure is useful for the calibration (in other words, scaling) of the bone so that an accurate

estimation of structural parameters follows. A number of studies has been focalized on search

for the construction a phantom, used for an efficient calibration of t bone specimens.

The phantom is needed to contain a number of physical elements of known geometries and

thicknesses to permit a number of measurements within a single micro-CT scan. It could be

stated that a micro-CT phantom should contain structures of material, shape and dimensions

similar to those of the usually examined objects by micro-CT scanning technique (e.t.c.

trabecular bone). According to the literature, the bone biopsies typically examined have the

shape of cubes or cylinders, with a size length or diameter (of 10-15mm order)(124, 45, 88 ). The

structure of a human trabecular bone could be represented as a mixture of rods and plates(105,

34) having mean thicknesses that can range from 100μm to 300μm (45). Consequently, the

thicknesses of the included structures have to be in this range. It has been considered

aluminum as an easy-to-handle material with an attenuation (μ) similar to that of bone.

Aluminum has already been used for calibration scopes in radiology (μAl(30keV)=3.04cm-1,

μcortical bonne(30keV)=2.56cm-1).(50, 9)

Aluminium has been chosen as an appropriate material in the design of phantom useful for

the calibration of bone samples.(104) Because the phantom is often used for quantitative

measurements, its contained elements have to be of previously established morphologies and

dimensions. The introduction of plate-like, rod-like and sphere-like geometries of known size

in the phantom is useful to control the assessment of the thickness in 3 dimensions and that

of the structure model index (SMI). The SMI is a topological index that gives an estimate of

the characteristic form (in terns of plates and rods) of which the bone structure is composed.

From a practical point of view, a small size of a calibration phantom (13mm diameter, 23mm

height) (104) allows its use with commonly used micro-CT systems, which can be of

commercial type, custom made or based on synchrotron sources.(97,120) According to

literature, the case of thin stuctures included in a phantom, being embedded in

polymethylmethacrylate (PMMA) allows the phantom to be applied for periodic measurementd

to monitor the performance of the mcto-CT system scanning bone samples.(104) It could be

added as a notice that a special application-dedicated phantom is certainly the most suitable

depending on the particular micro-CT examination and the type of measurement involved.(104)

3.1 LITERATURE STUDIES EVALUATING RELATIONSHIPS BETWEEN BONE

PHYSICAL PROPERTIES WITH MICRO-CT TOMOGRAPHY

This sub-chapter is dealt with the measurement of the structural parameters characterizing

the trabecular bone material using destructive techniques. An overall description of

assessment methods of evaluation of the paramteters defining the architecture (e.t.c. bone

volume fraction, surface area), density and mineralization process of the bone being research

object of publisized studies is given on this sub-chapter. In the content of this sub-chapter,

displays of the relationships between trabecular structural properties are presented.

PHYSICAL PARAMETERS (ARRANGEMENT OF) CHARACTERIZING BONE

STRUSTURE

It could be stated clearly that a comprehension of the architecture of trabecular structures

both in physiological formation and mechanical function is of main significance in the areas

dealt with prevention and treatment of bone diseases.(34) Concerning a descriptive study of

the trabececular bone structure(130), .the simplest kind of cancellous bone consists of

randomly oriented cylindrical struts (≈1mm diameter), each extending for about 1mm before

making a connection with one or more other struts, usually roughly at right angles. In a

variation of this pattern the cylindrical struts are replaced by little plates. The amount of

differentiation ranges cancellous bone in which there is

just the occasional plate among the struts to cancellous

bone in which there is just the occasional strut among the

plates.In other cancellous bone structure, the plates may

be considerably longer (several millimetres).

When this happers these longer plates are not randomly

orientated but are preferentially aligned in one direction.(27)

The final form of such cancellous bone..fine struts joining

them. Thus, it could be said that the characteristic form of

which the trabecular structure is composed could be more

plate-like, rod-like or even sphere-like trabeculae.

There has been created a number of physical models(34,

131) to simulate trabeulae structure (rod-trabeculae and plate-trabeculae forms) that provide a

means of analysing the structures through the use of simplified representations. (34) By

suitable observation and judgement, the trabeculae structure might be reduced o a physical

form retaining only the relavant aspects of the original and providing a physical model. (34, 131)

While this should be a subjective process, the validity of such a physical model may be tested

by formulating mathematical models from it and assessing their ability to predict the relavant

parameters. Physical models provide a basis for the promotion and verification of

conceptual(104,36, 35) models that explain physiological processes of bone remodelling in health

and disease.

It needs to be said that the description of trabecular structure has been achieved by several

physical modelling(34, 131), other mathematic formulations (40, 20). Concerning experimental

techniques used for the analysis of the trabecular bone material, it has been studied using

microtomographic techniques(38), histologic and histomorphometric techniques(104).

3.1.I BONE SURFACE AND BONE VOLUME MEASUREMENTS

It could be considered that the standard method of obtaining geometrical parameters of three

dimensional (3-D) structures is by applying stereological transformations to data obtained

from two-dimensional sections. (34) On this way, the mineralized bone volume per unit (Vυ) and

the surface area of mineralized bone volume per unit (Sυ) (called Specific surface by

Martins(0))might be estimated.

By Martin, (0) the most important features of the internal structure of bone are porosity and

specific surface of bone. Porosity could be determined by Martin(0) as the ratio of void volume

to the unit volume of whole bone or as the fractional part of bone occupied by soft tissues.

Figure 3.1.I.A: Image of human

femoral neck (There are many

plates and struts l

ortho

onal

to them) (140)

According to this study(0) bone is described as an inert building which is subjected to

substantial changes to its internal structure and composition as time goes by. These features

of bone are subjected to large differences by individual by Martin.(0) As it is said by on this

study, such internal changes and variations are wrought by prhysiologic processes which may

be classified as normal or diseased, but in either case occur on the internal surfaces of the

bone matrix (solid part of the bone material), or the walls of the voids depending on each

researcher’s point of view. Also, it is referred on this study that bone can be added on these

surfaces (by cells called osteoblasts) or removed (by cells called osteoclasts) from the inside

of the skeleton. Thus, it is pointed out by Martin() that the rate of change of porosity will be

influenced by the amount of internal surface that it is available for physiologic activity and thus

it is comprehensible that the porosity determines for a large extent the ability of bone to

function.

It is necessary to be said that a single measurement of Sυ and Vυ does not identify a particular

bone structure in any detail. However the way in which Sυ changes in relation to Vυ as the

structure remodels does.(34) According to the literature, the mathematical models have been

produced by the formulation of the dynamic relationships between Sυ and Vυ, the relationships

being inferred from the physical models by specification on the way in which remodelling of

the structures occurs.(34) It is referred on Fazzalari’s study(34) that bone remodelling comprises

the two processes of resorption and formation, both of which happen in discrete units of the

bone surface throughout the bone volume. It is additionally referred there that formation of

bone is coupled closely to resorption both in space and time, the combined process being

referred to as a remodelling unit. In addition, any long-term inbalance between formation and

resorption results in changes in bone mass. On Fazzalari’s study, (34) it has been assumed

that that the above changes of bone mass or equivalently that the remodelling units are

randomly distributed over the bone surface perpetually.

On Fazzalari’s study,(34) the two following processes were used in the creation of physical

models describing the trabecular bone functions. The trabecular region of interest was

considered to be composed of a fundamental architectural unit that specifically repeats to

produce the whole. (34) Thus, the dominant features of the structure were abstracted and

modelled within a unit cell. On Fazzalari’s study,(34) a series of models using the modelling

elements of plates and rods as well as model “plates/rods” (placement of a plate and a square

bar in unit cell arranged in parallel alignment) were created.

Fazzalari et al. (34) formulated a small range of possible physical models and they did not

model the commonly observed phenomenon of plates with perforaions as well as they did not

use modelling element distributed over a range of trabecular dimensions. Despite these

limitations, they verified that the uniform los of trabecular bone has a major impact on surface

availability for osteoblastic and osteoclastic activity and due to this situation the simple plate

model consisted of plates and rods is regarded as a limit for the analysis of trabecular

structure.

Anoher study(40) used a series of mathematical formulations for the examination of mean

intercept length applied on processing reconstructed micro-CT scanning data in order to

calculate the surface-to-volume ratio using human cancellous bone samples. The mean

intercept length(40) has been widely used in the field of biomechanics as a means for

predicting cancellous bone stiffness and strength using both empirical and highly theoretical

approaches. It is defined as the average distance between the intersections of the grid lines

placed over a histological section of bone with the bone-marrow interface. (40) The major

results of this study were the following referred. The theoretical estimations of the surface

distribution and average volume of the basic structural element of the cancelllous bone might

facilitate developing morphologically valid mechanical models for cancellous bone. Iit was

convinced that the strength of bone could be related to the structural element volume. Also, it

was verified that empirical relationships of this study that existed between the structural

element surface, structural element volume and bone volume fraction (BV/TV) were

remarkably potent and thus it was indicated that some underlying biological rules govern the

structural element size and structural element surface in cancellous bone.

A three-dimensional micro-computed tomographic study(38) dealt with the measurement of the

three-dimensional distribution of bone surface and the bone volume fraction

()

TVBV /of

human vertebral cancellous bone specimens.

An one-parameter nonlinear model explained a strong relation between TVBS /and

TVBV /. The ratio TVBS /in this model declined with thickening of trabecular plates and

rods which was convinced by the Fazzalari’s study.(34) It was convinced that TVBV /was a

good predictor of bone surface per total volume

(

)

TVBS /by using a one-parameter

nonlinear model

(

)

92.0

2=r.(38) A strong relationship between the TVBS /and

TVBV /was noticeable. (38) A theoretical independence between TVBS /and

TVBV /implied that biological mechanisms should exist which force the surface distribution

be related with TVBV /.

A three-dimensional simulation of bone remodeling is formulation on a more recent study(20).

This simulation takes into account the hierarchical structure of bone. The process of bone

tissue adaptation is mathematically described with respect to functional demands, both

mechanical and biological to obtain the bone apparent density distribution (at macroscale)

and the trabecular structure (at microscale). At macroscale, bone was assumed as a

continuum material characterized by homogenized mechanical properties. At microscale

(local scale) a periodic cellular material model approached bone trabecular anisotropy as well

as bone surface area density.

As it is referred on study,(20) a morphometric parameter describing trabecular architecture is

the surface-to-volume ratio known as “bone surface area density”. Several studies(34, 20, 40, 38)

as reported above have studied the surface-to-volume ratio for for both bone samples and

bone idealized microstructure models and concluded that a strong relationship exists between

surface area and apparent density.

Figure 3.1.1.B illustrates that the bone surface area avail able for remodelling is not a uniform

function of density but rather has a maximum at bone intermediate volume fraction values.

3.1.II ASH DENSITY VERSUS BV/TV BONE DENSITIES (APPARENT, TISSUE DENSITY)

Density is considered to be a main determinant of the mechanical properties of cortical and

cancellous bone tissues.(140) Density is defined in a number of ways at either the bone tissue

(app

Dapparent) or the bone material level ( mat

D material).

Ash fraction (or mineral content) is rarely presented in studies, but it has recently received

increased attention in the literature. Variations in ash fraction can be caused by bone

diseases (osteomalacia, Paget’s disease) or treatment with certain antiresorptive drugs

(bisphosphonates).(45)

Mineral content of bone has been related to the volume fraction and the apparent density of

the trabecular bone structure. (140) The bone volume fraction is a geometric and visible marker

of the level of the porosity in the structure.

It has obtained from a study(45) that bone volume fraction and ash fraction are poorly

correlated which observation is opposed to the study(131) where a linear regression between

ash density and bone volume fraction showed a high and significant correlation for trabecular

and cortical specimens as well as another study by Schileo(128.) where an excellent correlation

was convinced for the pooled cortical and trabecular specimens by ρash/ρapp linear

regressions.

As it has been reported in the literature, mechanical testing of bone samples has identified the

apparent density and the ash density as effective predictors of bone strength and stiffness.

On the following text, the relations between ash density and bone volume fraction as well as

the other types of bone densities are describes according to results of different studies.

Hernandez’s study (45) having as a primary objective the evaluation of the relative influence of

volume and ash fraction on bone strength and elastic modulus over a broad range of volume

and ash fraction values, convinced a poor correlation between bone volume fraction and ash

fraction

(

)

01.0

2=r(Figure 3.1.II).

Hernandez’s study() used specimens tested by Keller(64).

Keller’s data were obtained from spinal cancellous bone and femoral/diaphyseal bone

specimens. As it is displayed by (Figure 3.1.II) the distribution in bone volume fraction and

ash fraction is larger than in previous studies (as it is seen from Table 3.1.II).

Study(140) based on search that took place in Biomechanical laboratories (Cranfield University,

Defence Academy , UK) convinced a bimodal relation between Mineral content

()

%Min and

(

)

BV due to a proven tight relationship between Dmat and

(

)

%Min .(at it is displayed by

Figure 2.2.4 on page 23 of this study)(140) As it could be showed up by Figure 2.2.4 the lines

composing the figure denote to this an envelope form drawn by hand. By obserbing Figure

2.2.4 it could be claimed that specimens characterized by Dapp<1.3 gcm-3 presented a

decreasing linear behaviour of mineral content as a function of BV/TV in comparison with

them being characterized by Dapp>1.3 gcm-3 that presented an increasing linear behavior of

mineral content related to BV/TV. It needs to be said that bone specimens had been removed

from bone marrow and fat before the measurements.

An experimental-numerical Schileo’s study (128) was performed to investigate the relationships

between computed tomography (CT)-density and ash density and between ash density and

apparent density for bone tissue so that it evaluates their influence on the accuracy of

subject-specfic Finite Element Models of human bones. During the prerformance of this

search CT-densities of specimens were computed of CT images while apparent and ash

densities of human and bovine femurs were measured experimentally. Due to the important of

the derivation of the bone specimens for the calculation of bone densities, it would be

necessary to point out that cortical specimens were removed from the diaphyses and

trabecular specimens from the epiphyses of bovine femurs.

Concerning the preparation of the specimens prior to measurement of apparent density, the

most important to notice is that bovine (cortical and trabecular) and human femoral trabecular

specimens(88) were washed efficiently so that the bone marrow be totally removed and the

excessive water was removed from the marrow cavities.

The ash density versus apparent density plots are displayed on Figure 3.1.IIb.

Characterizing the plots described above and thus referring to the ρash/ρapp relationships, it is

resulted that the average ρash/ρapp ratio for trabecular bone was 0.46, ranging from 0.34 up to

0.62. (128) The ratio seemed to decrease as tissue density increased but this was rejected on

the basis of the results obtained from thehuman femur trabecular specimens. (128) Additionally,

by looking on the above plots (Figure 3.1.IIb), it can be noted by researchers. (128) that ρapp

can be precisely measured in low-density trabecular bone specimens in low-density

trabecular bone specimens, but an overestimation of the ρapp of large-size high density

trabecular specimens, where it is difficult to be certified the complete removal of marrow and

water from the inner cavities.

Tassani’s study(131) had as an objective investigating whether tissue mineral density could be

assumed as a constant in adult human (trabecular and cortical) bone. On this study, an inter-

site analysis was performed on cortical and trabecular specimens extracted from different

anatomical sites and an intra-site analysis was performed on specimens extracted from

femoral heads. Bone volume fraction was computed by mico-tomography technique.

Concerning the specimens used in inter-site analysis, cortical specimens were extracted from

the diaphyses of tibias and femur while trabecular specimens were extracted from their

epiphyses. (131) Specimens used in intra-site analysis were extracted from different femoral

heads.(131)

Figure 3.1.IIB showed up a single linear regression between ρash and BV/TV for the in both

particular and cortical tissues. Therefore Tissue Mineral density (TMD) was supposed as a

constant as it was defined as the ratio between ρash and BV/TV. TMD (Figure 3.1.IIC) was

found to be relatively constant in the whole range of BV/TV. (131) As it is written on Tassani’s

strudy(131), no significant difference was found beteen trabecular and cortical TMD.

Tassani’s study(131) is coming on disagreement with study(140) based on search that took place

in Biomechanical laboratories (Cranfield University, Defence Academy , UK). Zioupos’

study(140) convinced a “boomerang” like distribution for both mineral content and material

density. As it is supported on Tassani’s study no comparison could exist between these

studies because material bone density was not examined on his study. Tassani’s study(131)

can not verify the correlation between bone mineral content and material density that was

convinced Zioupos’ study(140) because Tassani(140) did not measure organic part of tissue.

Differences between the bone mineral content between these studies and Schileo’s(128) can

be placed in the cleaning of the small pores of trabecular bone or removing excess water and

marrow of the pores existed in trabecular bone. There are existed differences in displaying

the plot of Ash density versus BV/TV between Tassani’s study(131) (Figure 3.1.IIA) and

Zioupos’ study (Figure 2.2.4). They can explained by the following reason. The specimens in

Zioupos’ study(140) have been extracted from the whole of the bone and so they are

characterized by a varied range of BV/TV and Ash content. However, specimens has been

removed only from the epiphysis and diaphysis in Tassani’s study(131) . The BV/TV and Ash

content that characterise specimens in Tassani’s study(131) have not depended on the same

way as Zioupos’ study(140) because they may have not been extractred from regions of highly

varied BV/TV and Mineral content (Ash content).

3.2 LITTERATURE STUDIES EVALUATING RELATIONSHIPS BETWEEN BONE

MECHANICAL AND PHYSICAL PROPERTIES WITH MICRO-CT TOMOGRAPHY

The mechanical properties of bone are affected greatly by the porosity, the mineralization

level and the organization of the solid matrix.(112) It has been referred on studies searching for

the mechanical properties of bone that beyond relying on average mechanical properties of

bone, better approximations of bone elasticity have been obtained through correlation with

density.(16) Indeed, it is convinced in some studies, that direct measurements of types of

density such as apparent hydrated density, apparent dry density had been correlated with the

elastic modulus and strength of bone. (16, 76)

This sub-chapter is dealt with the overview of publicized studies which investigated the

correlation between physical and mechanical properties chatacterizing the bone structure.

CORRELATION OF MECHANICAL PROPERTIES TO PHYSICAL PROPERTIES OF BONE

STRUCTURE

There have been numerous studies in literature that examined the modulus-density and

strength-density relations on the bone structures.(65, 115, 57) Researchers(65,115,57) dealt with the

establishment of these relationships by measuring density of bone specimens using

histological examinations and micro-CT analysis and assessing mechanical properties (elastic

modulus, ultimate strength) with mechanical experiments.

Being referred to a number of the studies examining mechanical properties as a function of

density it could be said that Keyak(65) examined the possible modulus-density relationships

and strength-density relations in order to verify corresponded results of previous studies.(16, 76)

This study dealt with the proposal of relationships between mechanical properties and density

by using human trabecular specimens derived from proximal tibia. The ash density of the

trabecular bone specimens was determined after their ashing in a muffle furnace at 600oC for

24h and their dry density was evaluated after their air drying for 24h. Their mechanical

properties (elastic modulus, ultimate strength) in different directions were obtained from

mechanical tests.

A majority of literature studies publicised earlier than Keyak’s(65) examined elastic modulus

and strength of bone in the Superior-Inferior (S-I) direction (Figure 3.2.I,, Figure 3.2.II). By

Keyak’s study(65) a S-I modulus-dnsity correlation indicated a greater modulus at each density

than all the correlations presented in other studies except for Ashman et al.(2) This differenced

can be expained lying several factors. Linde(76) and Keaveny (61) showed up that increasing

the specimen size and length-to-width ratio tended to rise the measured elastic modulus of

trabecular bone. Keaveny (61) found that the elastic modulus of 5-mm cubic specimens were

36% higher than those of 10-mm long by 5-mm cylinder specimens. In Keyak’s study(65) 15-

mm cubic specimens were used.

Effects of specimen geometry on strength have been publicised by Keaveny (61) who

convinced that their bovine cube specimens were 18% stronger than their cylindrical

specimens. Keyak’s results(65)are in accordance to those of previous studies despite different

geometries (Figure 3.2.II), with the exception of Odgaard() who used the most slender

specimens.

The strength and elastic modulus value measured in Keyak’s study(65) may have been

understated because multiple tests were performed on each specimen. Despite Linde’s(78)

findings implied that repeated orthogonal testing to 0.5% strain had not leaded to damage,

Keyak(65) found the modulus to be significantly lower during destructive tests than during non

destructive tests performed before destructive tests. Additionally, it could be said about

Keyak’s study(65) that the higher maximum strain (0.67%) in conjuction with the multiple tests

having been performed, might had leaded to fatigue damage in Keyak’s(65) specimens,

resulting in lower modulus and possibly lower strengths.

Rho(115) studied mechanical properties of cortical and cancellous bone specimens by using an

ultrasonic transmission technique. Cortical bone specimens were extracted from human tibia,

femur , humerus and mandible whereas cancellous bone specimens were extracted from

proximal tibia, proximal femur, distal femur, proximal humerus, patella and lumbar spine. Raw

computerized tomography (CT) values that obtained from scans of bones in water were

corrected in Hounsefield units (HU).

The density of cortical bone was determined by Archimede’s principle(115) while the apparent

density of cancellous bone was determined by histological technique. Bone marrow was

removed from cancellous bone specimens.

It was proven by Rho’s study(115) that axial modulus was predicted by density of cortical bone

on a best way through linear rrelationshsips. He found that a power function of densiy can

predict the axial modulus similarly as a linear function. However, he found that power models

between elastic modulus and apparent density produce statistically better fits. Indeed, he

found that the power function gives a better fit of data at the low and high densities of

trabecular bone.(Figure 3.2.II.A)

3.3 LITERATURE STUDIES EVALUATING THE ASSESSMENT OF MECHANICAL

PROPERTIES OF BONE BASED ON COMBINED USE OF COMPUTED TOMOGRAPHY

AND FINITE ELEMENT MODEL TECHNIQUES

The mechanical response of an individual patient’s bone and the proximal femur in particular

is of major clinical importance for orthopedists.(145)In specific, predicting the mechanical

response of the proximal femur for individuals is of major clinical importance as a planning

and analysis tool to assist orthopaedics in fracture treatment planning.(146) This prediction can

help surgeons determine whether a surgical or non-surgical treatment is preferable and when

the treatment is identical to choose the optimal implant type, size and implant/screw position.

Similuation of the mechanical response of a bone (e.t.c. proximal femur) is nowadays limited

because it depends on the acquisition of bone’s exact complex geometry and its anisotropic

and inhomogenous material properties that vary among individuals.

Last decade-advances of quantitative computerized tomography (QCT) and high-order finite

element methods (FEM) make possible performing of reliable patient-specific mechanical

simulations of loading conditions applied on the proximal femur(145) For instance, QCT scan

data has been used for the creation of accurate FE models of the femur.(65, 20, 88, 140, 26)

Yosibash study published in 2006(145) utilised QCT scan data for the generation of a precise

geometrical representation of an individual bone that incorporates a model of the internal

surfaces that separate distinct trabecular and cortical regions in it. Moreover, QCT information

can be correlated to the local density of bone so that providing of inhomogenous, region-

specific distributions of the density be obtained within the bone which can possibly further

used in order to determine a functional distribution of Young’s modulus.

Except for QCT, the evolution of micro-computed tomography (micro-CT) was initially driven

by the need for providing (having) a highly precise and accurate means of reconstructing the

complex architecture of bone tissue at a high resolution (95, 92). Micro-CT plays a determinant

role in the evaluation of disease pathogenesis(10) and efficacy of interventions(123). Micro-CT

scanning can be achieved at resolutions as low as 5μm, allowing the determination of

porosities. Futhermore, original three dimensional (3-D) image reconstructions permit the

assessment of bone microarchitecture as a 3-D structure, adding critical information to

images collected through histomorphometry, which represents only bone in unconnected

slices.(92)

Utility of micro-CT scanning is extended to the generation of micromechanical models of the

scanned tissue by directly converting bone voxels to mechanical elements that can be

analyzed using the FEM. Through the voxel-conversion technique, bone voxels are converted

to equally shaped brick elements such that the resulting finite element (FE) model has exactly

the same geometry as the reconstruction in its derived pattern.(10, 108, 135, 118) Finite element

modeling has been successfully implemented to model the bone from the mechanical aspect,

enabling the simulation of the mechanical behavior of the scanned voxel structures in silico.

On a similar way to that of a mechanical test, Finite element modeling can facilitate the

calculation of the overall mechanical properties of the bone. It needs to be reported that the

FE software can model a mechanical compression test on a specimen, assessing the

apparent stiffness as well as apparent stess-strain parameters under a loading condition.

This sub-unity is focalized on the decription of the analytic process of deriving the mechanical

properties of bone from CT-scanning images. This process in counted on the implementation

of the finite element modeling on the reconstructed CT-scanning images intending to

determining (evaluating) mechanical behavior of bone tissue,

3.3.I.SEGMENTATION AND CONTOUR EXTRACTION

Segmentation process is carried out on images obtained from micro-computed tomography

(micro-CT) in order three-dimensional geometries of bone tissue are created.(17) Segmentation

is the procedure that basically divides the image in parts having similar characteristics. In

literature studies, it is referred that segmentation procedure separates the bone from the

marrow, other soft tissues and background (6, 17) surrounding bone in micro-CT scan images

by using the binerization of images.(6)

It is evidenced by literature studies that specialized segmentation packages (such as

VOXELMAN (90)) have been used for the segmentation of 3-D micro-CT scanning images.

Software packages as VOXELMAN for example can possibly visualize three orthogonal views

of a 3-D data set of scanning images as well as accomplish the creation of a histogram-based

thresholding.(9) The segmented images can be rendered in three dimensions after the

application of several morphological operations and a subsequent connected component

analysis.(9) There software packages as VOXELMAN(90) that allow the user to interactively

remove any counterfeit labels aiming to obtain an accurate segmentation. It needs to be said

that this this type of segmentation is tedious since it demands a large amount of human

interaction.(9) Due to the need for an automatic procedure, it appears to think considerably of

the materials to be segmented.(9) Indeed, due to characterization of bone as an

inhomogenous material (Figure 3.3.I.A), the selection of the threshold value on a reliable way

could be considered as impossible.

As a following of the previously reported procedure, the performance of edge detectors(14)

which detected and placed sharp boundaries for the automation of the segmentation process,

was assessed on a study.(9) However, then the texture created a large number of fake edges

that made this approach unreliable. In the contrary, an efficient region-based algorithm based

on advanced morphological functions yielded precise results. On a next step, a grey-scale

dilation on a large neighborhood followed by a grey-scale reconstruction(140) can facilitate the

removal of the texture and the creation of two separate regions.(Figure 3.3.I.B). The

histogram based on a grey-scale imaged reconstruction is bimodal, which infers that a

threshold value can be possibly chosen that produces an initial coarse segmentation.(Figure

3.3.I.C). It is referred by Bossart(9) that the extraction of watershed lines uses as inputs both

the initial coarse segmentation and the gradient stage which yields an accurate solution being

displayed on (Figure 3.3.I.C).

Occasionally, the forms of reconstructed images that have been produced on the end of the

above referred procedure have to be subjected to the influence of smoothing filter. It would be

considered as necessary mostly on the cases of reducing the sharpness of grey-level

reconstructed images.

3.3.II ASSIGNMENT OF MECHANICAL PROPERTIES OF BONE USING FINITE ELEMENT

MODELS

GENERATION OF FINITE ELEMENT MESHES

According to the unity 3.3.I, exterior interface and interior boundaties and traced(145) and x-y

coordination systems are generated, where each of them represent different boundaries of a

defined slice (two left images (intermediate, outer boundary detection) of Figure 3.3.II.A). The

exterior, interface and interior boundaries are often influenced by three-dimensional (3-D)

smoothing algorithms who affect by reducing the sharpness of them. By (Figure 3.3.II.A)(9)

the quality of the initial exterior, interface and interior boundaries is influenced by a 3-D

spherical filter, a two-dimensional (2-D) averafing filter and a spline interpolaton (where a 3D

spherical filter calculates the new location of each point in a specific slice using data from

other slices around it). The smooth edges using cubic spline interpolation are saved in form of

array Smooth edges that can be read in a CAD package.(145) After the generation of splines,

the surfaces and subsequently solid bodies are generated. Then, each one of the generated

3-D solids is imported into a p-Finite Element (FE) solver and a mesh is generated by an

auto-mesher using tetrahedral(on) elements, as well as hexahedral elements. The elements

follow exactly the surfaces of the bone(145).

ASSIGNMENT OF MATERIAL AND MECHANICAL PROPERTIES

After the generation of meshes, each element of the mesh is given a different Elastic modulus

(Young’s modulus) calculated as a function of the density information (data) which is derived

from Computed Tomography (CT) dataset. An average Hounsefield unit (HU) is computed for

each element through a numerical integration of the HU field over the element’s volume.

A linear relationship between the HU numbers (Figure 3.3.II.B) and the bone density can be

assumed based on different studies. The CT dataset is calibrated using a calibration phantom

with bone-equivalent (solution of hydroxyapatite) insertions of different densities which is

placed as close as possible to the bone in order that errors introduced by non-uniformity of

CT-numbers within the scan files, are minimized.(145) It is commonly assumed that a power

relationship exists beteen Young’s modulus and bone density (e.t.c. ash density).(64). A

constant Poisson’s ratio value is often assumed. By solving a set of mathematical equations

that are formulated as an application of Hooke’s law we can feasibly estimate the overall

(apparent) mechanical properties of a bone structure.

3.4 FACTORS AFFECTING MEASUREMENTS OF MECHANICAL PROPERTIES BASED

ON THE MICRO-COMPUTED TOMOGRAPHY AND ON FINITE ELEMENT METHODS

This sub-unity is referred to a brief description of the main factors that influence

influences the quality of the CT-scan imaging and thus the values of the physical and

mechanical properties of determined based on CT-scan reconstructed images.

LITERATURE REVIEW

According to a research on the literature for studies which examine factors that affect possibly

the quality of micro-computerized (micro-CT) scan images and thus the estimated properties

(material, mechanical) of the bone tissue, our interest is focalised on numerous important

studies.(67, 24, 110, 86, 71, 148) Firstly, it could be said that Kim, 2004(67) evaluated the effect of

scanning and reconstruction voxel size on the micro-CT based stereological analyses of

human trabecular bone. He had intended to quantify the variations of morphological

parameters at different voxel sizes. By accomplishing his measurements, he found out that all

sterological parameters (BV/TV, Tb.N, Tb.Th, Tb.Sp, BS/BV) varied with different

scanning/reconstruction voxel size groups. (Table 3.4.I) After the accomplishment of a two-

way repeated measures ANOVA, it was suggested that scanning voxel size is the major

factor causing the differences of values in BV/TV, whereas reconstruction voxel size did not

effect significantly BV/TV. Concerning Trabecular Number (Tb.N) and Trabecular Space

(Tb.Sp) significant differences were observed due to reconstruction voxel size, but not due to

scanning voxel size. The differences that were noticed in Trabecular Thickness (Tb.Th) and

Bone surface-to-volume ratio (BS/BV) were significantly related with the reconstruction and

voxel size (Table 3.4.I)(67).

Study achieved by Cooper(24) examined the impact of voxel size on 3D micro CT analysis of

trabecular and cortical bone morphological parameters. It was shown by Cooper’s

measurements that the highest scan magnification ( (5μm, 10μm, 15μm) scan size, (10μm,

20μm, 40 μm) voxel size) provided the sharpest detail and as voxel size increased the images

become progressively distorted.(Figure 3.4.I)). The 3D renderings demonstrated the same

general pattern with decreased detail.

The dependency of morphological parameters of cortical and trabecular bone on the

resolution/ voxel size is displayed on (Table 3.4.II) in Cooper’s study(24). It can be observed

that mean trabecular thickness seemd to increase as voxel size increased, as it is proven by

the previous studies. This can be attricuted to the loss of detection of smaller trabeculae

resulting in evaluated mean thickness(69). Merging of trabeculae due to an increasing in their

thickness can possibly account for the decrease in trabecular number associated with larger

voxel size.(67) In Cooper’s study(24), canal number decreased with magnification for the scan

and the artificial datasets. While a merging effect can not be rule out out for this study (24), it

would likely affect slightly the measurement of cortical canals which then, they have a more

spatially arrangement than the plates and struts of trabecular bone. The decrease in he

cortical canal values is fundamentally attributed to the loss of detection of small canals and

thus the subsequent rise in the space between those that remained. On a similar way, the

increase in the mean trabecular separation (Tb.Sp) due to increase of magnification. (Table

3.4.II) BV/TV of trabecular bone had been steadily decreased as it is described in studies

which investigated thin on magnification scale with voxel sizes below 100μm. (Table 3.4.II)

Another study done by Rajapakse(110) investigated the accuracy of mechanical parameters

obtained by Bone Volume Fraction (BVF) scaled micro Magnetic Resonance Imaging (μMRI)-

based micro Finite Element (μFE) analysis of bone at resolutions achievable in vivo by

comparison to those derived from high-resolution reference μCT images of human distal tibia

specimens. It was proven(110) that increased voxel size causes systematic overestimation of

mechanical parameters compared to the reference values (Figure 3.4.II) resulting from

artificial connections being introduced into the Trabecular Bone (TB) network once voxel

dimensions approach and exceed those of trabecular thickness and spacing.

However, the correlation between mechanical parameters derived on the basis of simulated

low-resolution images remained strong compared to the reference high-resolution (25μm)

values even at voxel sizes (160μm) accomplishable by in vivo μMRI.(110) It is useful to be

noticeable that elastic modulus (in third (superior-inferior) direction) was laregely determined

by the trabeculae along the principal loading direction of distal tibia , thus within limits lowing

resolution along this direction will not influence the recovery of these trabecular.(110)

Meganck(86) studied the beam hardening artefacts associated with micro CT imaging. He

found that filtration of the X-ray beam does not demand software based on beam hardening

corrections, but this decreases contrast, as well as decreases the baseline noise and

decrease throughput. By Meganck’s study(86) beam hardening induces less artifact for

morphology than densitometric measurements.

Kyriakou(71) proposed a dedicated image-based ring artifact correction method for high-

resolution micro CT based on median filtering of the reconstructed image and working on a

transformed version of the reconstructed images in polar coordinates.

By Kyriakou’s study(71) ring artifacts are developed by imperfect detector systems that

significantly under- or overestimate attenuation values throughout the scan process.

In his study, ring artifacts were shown to increase noise and inhomogeneity in CT images

which can possibly result into a reduction of low-contrast resolution and potential

disadvantages are existed with respect to diagnostic results in the area of pre-clinical imaging

and the quality decreases more when central detectors are affected by creating a dark

smudge at the center of the image.(6)

According to Sheng’s study(148) the center of rotation is a physical location in the

microCT scanner, defined by the axis of rotation of the sample stage. This physical location is

always well defined during calibration of the instrument and this is fitted(148) by an appropriate

algorithm. However, in real images of limited contrast and with the influence of X-ray photon

noise, this algorithm exhibits poorer precision and the optimum center of rotation cannot be

always acquired. Thus, adjustment by operator is necessary to determine whether the center

of rotation was correct, in order that the structural information of the sample can be correctly

interpreted.

Table 3.4.II shows the varied values of material and structural properties being influenced by

the deflection of the center of rotation. On Sheng’s study(148)Tissue BMC and BMD were

measured, based on the threshold, which was used to distinguish bone tissue from marrow.

The number of voxels representing bone structure increased as the automatic threshold

decreased with the center displacement.(,71) Although tissue BMD is defined as the ratio of

tissue BMC to bone volume, it did not increase with the increase in tissue BMC or bone

volume.(71) However, it decreased with center displacement. This indicated that bone volume

rather than tissue BMC was more easily influenced by center variation.(148) A bone

microstructural analysis revealed that Tb.N and Tb.Sp were not influenced by the center

displacement.(148) Tb.Th was more sensitive to changes of center than Tb.Sp although the

same algorithm of direct measurement was used for both parameters.

BS/BV decreased with the center deflection. This finding was consistent with those of

previous studies in which an increase in both Tb.Th and BV/TV had been found,

accompanied by a decreasing BS/BV.(67) A relative decrease in the bone surface due to

trabeculae thickening and an increase in bone volume contributed to the decrease in BS/BV.

The unchanged (Structural Model Index) SMI and Degree of Anisotroopy (DA) indicated that

although the distortions and the associated loss of spatial resolution that was particularly

evident in samples having plate-like structures,(57) the volume ratio between plates and rods

and the preferred orientation in trabeculae was relatively intact. (SMI) was a parameter which

used to evaluate the plate-rod characteristics of the structure.(30)

CHAPTER 4: MICRO-CT SCANNING ANALYSIS OF BONES

On this short chapter, a process followed during an applied micro-CT scanning analysis on

tibia specimens is described. The preparation procedure of the samples for the micro-CT

scanning analysis is described which is followed by a description of their scanning analysis

using micro-CT scanner.

SAMPLES USED FOR MICRO-CT SCANNING MEASUREMENTS AND SAMPLE

PREPARATION

4.1 SAMPLE PREPARATION

Animal treatment was done during the previous experiment. (38) The study protocol was

reviewed and approved by the Animal Experiment Committee of the University of Kuopio

(Kuopio, Finland). Fifteen samples treated with two doses (control and 1µg/kg) were packed

into resin during previous experiment measurements.(38) Three other doses, 0.03µg/kg,

0.1µg/kg and 0.3µg/kg, were embedded into resin in that project. Each dose was prodided to

seven samples.

Tibias of the three doses were dissected clean of muscle and they were remained in dried

condition, in the oven (37°C) for a week before treatment. Bones were held in position with

plastic frames and set vertically in the cylindrical potting mould (inner diameter 32mm).

Polyester resin, which was made from 6 drops of hardener to 10ml or resin (MetPrep Ltd,

Kleer-Set Type FF, Coventry, UK), was added into the moulds such that the metaphyses of

the samples were covered by a minimum of 1cm. (Figure 4.1)

The samples were hardened at room temperature for a day. Draper 360mm (14”) Band saw

was used to create parallel sided blocks in order to increase the stability of holding the

samples in the clamps when using metallurgical saw (Struers, Accutim-2, Rodovre,

Denmark). Slices were cut at approximately 15% of the tibial length at the proximal end.

These slices were measured approximately 1.4mm in thickness. Slices from previous doses

(control and 1µg/kg) were soaped in distilled water for a day to re-wet the samples. All slices

were ground by using a 0.05µm γ-alumina slurry (MetPrep Ltd, Coventry, UK). The number of

samples treated with different doses is displayed on Table 4.1.(Edmun) Microscope was used

to see if the surface of the sample was smooth enough to be analyzed (without as many

scratches as possible). The slices were ultraonicated for three times of 30 seconds and left in

the oven (37°C) for further analysis.

4.2 DESCRIPTIONOF THE STUDY (Edmun Chun)

All the samples were scanned by using micro computed tomography (m-CT) analysis. The

micro-CT (XR H 225) scanner was used for the determination of the mineralization and

porosity of the bone. The scanning analysis of the bone samples was performed under the

accomplishment of the Msc thesis by Yiu Kan Chun.The radiation was generated by a micro-

focus source and this was transmitted through the sample. A digital flat panel detector was

used to capture patterns of X-rays that passed through a specimen by showing different

shades of gray depending on the material and the geometry of the scanned bone tissue. CT

analysis offers this additional benefitial aspect to the X-ray technology. Based on a large

number of X-ray images being captured around a single axis of rotation, CT analysis can

reconstruct a 3D volume dataset characterised of high accuracy at most, that represents the

internal structure of the sample.

4.3 MICRO-COMPUTED TOMOGRAPHY

Five standards were used throughout the scanning analysis. After the scanning through mCT,

the images were reconstructed by CT-pro software and attenuation coefficients were gained

using VGStudio Max 2.0 software. By plotting the relationship between the grey level and

mineral content of the standards, the mineral content of the samples could be obtained. Table

7 show the mineral content of five standards.

CHAPTER 5: PROCESS OF MICRO-CT SCANNING IMAGES

On this short chapter, the process of the micro-CT scan images by using CT-Pro software in

order to obtain the micro-CT scan images reconstructed is described through a detailed

description. The micro-CT scan images were obtained through the micro-CT analysis which

was described in the previous short-chapter.

The reconstrcuction of the micro-CT images that were produced during the micro-CT

scanning analysis that was described in previous short chapter (Chapter 4) will be made by

the re-calculation of the center of rotation. The process of re- calculation of the center of

rotation by using scanning images will be described here in detail. Followingly the process of

the re-constructed images with a re-calculated center of rotation through vglStudiomax

ssoftware will be described in detail intending to define the external contour and internal

surface of the bone and then assess parameters characterizing its structure is derived.

5.I CT-PRO SOFTWARE

5.1 OVERVIEW OF USER INTERFACE

CT-PRO software is useful because this performs the offline 3-D Computer Tomography

reconstruction . One possible way of using useful functions incorporated on this software

package is described on the following text.

By loading the CT-PRO software package in our system, selection and set up controls appear

on the left part of the interface (Figure 5.1.A) and image display window appears on the right

part of the interface.

Next, by using File Open (Alt+F, O) an open file dialog is displayed through of which we can

browse to the \CTData and select an *.xtekct parameter file to open. (This *.xtekct file has

beern created by *X.Inspect software in end of micro-CT scanning process.) If *.xtekct file is

valid, the first projection existing among the projects dataset of projections will be displayed

on the Image Tab (Figure 5.1.B).

5.2 PROCESSING TABS OF USER INTERFACE

The processing tabs that are ordered left to right (as it seems in the user interface) in order to

guide you through the process of loading, setting up and then producing a volume following

the correct order are appeared in the upper and right part of the user interface (Figure 5.3).

The function that each of the procession tabs perform is describes on the following text.

The Image tab examines the loaded dataset each time. User is helped through Image tab to

check that the sample has not been moved during the CT scan. If the object has been moved

during CT scan, this movement would occur due to insufficient support over the micro-CT

scanning of sample.

The Centre of Rotation tab is useful for the correction of centre of rotation due to its

misalignment with respect to the X-ray axis.

The Set up tab is activated for the beam hardening, noise reduction and other functions which

can possibly change the quality of the projection of the data set.

The Calibration tab is used in the case of calibrating of the reconstructed image for the setting

of scales (e.t.c. Hounsefield Unit (HU)) on this image.

The Volume tab is utilized for the provision of three dimensional (3-D) volume information for

a given region of interest and the resolution concerning the quality of the reconstructed

image.

5.3 RECONSTRUCTION OF THE SAMPLE USING MICRO-CT SCANNING IMAGES

The reconstructed image is produced by accepting as an input image files of the form

(*.xtekct file). This file is got through a complicated process until the reconstruction procedure.

This proceduce is described analytically below.

Throughout the reconstruction process, the calculation of center of rotation and the use of

Beam Hardening and Noise reduction function is acquired intending to get a high-quality

reconstructed image. In particular, the calculation of the center of rotaton is useful due to

possible misalignment with the X-axis.

Firstly, the Center of Rotation tab is selected (an analytical description of the process being

achieved by the center of rotation tab is described on 5.4 CALCULATION OF CENTER OF

ROTATION) aiming to the calculation of the center or rotation of micro-CT scan images. After

the calculation of the center of rotation, we type 'Start' in the bottom of the tab for another

time (‘Start’ has been typed at first for the beginning of the calculation of the center of

rotation)

in order to reconstruct a selected stack of images.

On a next stage, we go further the reconstructing procedure by selecting the

'Set Up' tab. “Beam Hardening” and “Noise Reduction” functions can be used

ideally intending to obtain a nice qualitative reconstructed image. These functions could be

used in a difference sequence on the “Set up” tab. By selecting the “Beam Hardening”

function, we have to type the “Reconstruction all” button presented on this level. Also, we

select to take a 'best resolution' recontructed image' in the 'Reconstruction of a slice' stage of

this tab and then, click on 'Start'. Next, by selecting the ‘Noise Reduction” function, we have to

do similarly as by selecting the “Beam Hardening” function. We have not to forget opting to

take a 'best resolution' recontructed image' in the 'Reconstruction of a slice' stage of

this tap and click on “Start”.

By following the above stages, we move the mouse on the 'Volume' tab, where we submit the

volume name (*.xtekct file) (in Volume Reconstruction frame') for the reconstruction process.

We have normally to be informed that CT-Agent software have already accepted our named

file aiming to set this in order for the reconstruction process.

On the last stage, we use the “Calibration tab” by selecting to implement the “Reconstruction”

on the slices that constitute the volume of the sample where we opt to acquire a ““Best”

quality & “100%”” reconstructed volume. If “CT-Pro” which collaborates with “CT-Agent’ during

the reconstruction process does not begin accomplishing this process we will have to check

the schedule of the reconstruction jobs. “CT-Agent” image is displayed on the toolbar which is

appeared on the bottom of our computer screen “ CT-Agent” plays an important role during

the reconstruction process because it receives each time the projection images and the

reconstruction parameter files for the reconstruction of the volume of the sample as well as

collaborating with ‘CT-Pro” for the accomplishment of the reconstruction process. If “CT-

Agent” is in the state which is characterised by the signal “Resource busy” any

reconstruction job will be paused and the Receiving files state can not be entered. The

reconstruction job will be re-started when “CT-Pro’ has finished reconstructing slices and the

“Resource Busy” state is exited (Table 5.I). When the stage of the reconstruction process is

completed through the “Calibration” tab we use the “Volume” tab for last time, where we

submit the volume name (*.xtekct file) (in Volume Reconstruction frame') for the

reconstruction process. Lastly, it is important for us to call our created reconstructed file

differently. Our created reconstructed file is of the form (*.vgi.file) and they will be introduced

in “vglStudiomax” software

5.4 CALCULATION OF CENTER OF ROTATION

DEFINITION OF THE CENTER OF ROTATION

According to Sheng’s study(148) the center of rotation is a physical location in the

microCT scanner, defined by the axis of rotation of the sample stage. This physical location is

always well defined during calibration of the instrument and this is fitted(148) by an appropriate

algorithm. However, in real images of limited contrast and with the influence of X-ray photon

noise, this algorithm exhibits poorer precision and the optimum center of rotation cannot be

always acquired.

The center of rotation misalignment of the object rotation point with the X-ray center axis

occurs due to the following reasons():

i) Thermal movement, which may sets the necessity of the installation of a cooler.

ii) Depression of the arm due to the weight of the sample

iii) Placing the sample on the turntable

iv) Not vertical position of the turntable due to the above reasons

v) Poor maintenance and calibration

EXPLANATION OF THE FUNCTION OF CORRECTING CENTER OF ROTATION WITH

CT-PRO

“CT-Pro” software determines the correction of the center of rotation that is necessary each

time. The sharpness of the data is assessed. Then, the projections are shifted by a pixel, the

reconstruction is re-done and the sharpness is re-assessed. On this stage, the required shift

is estimated and iterative movement and reconstruction is done until the slice has reached the

desired quality. This process is done for a slice near the top and bottom of the region of

interest and these two values indicate the misalignment at the top and bottom of the image

and thus the shift and shear are needed to correct all the projections. The correction at the

top and the bottom are quite similar and thus a single slice in the centre of interest is

sufficient. Nevertheless, using two slices at the top and bottom of the region of interest can

contribute in the derivation of a better result.

USING THE CENTER OF ROTATION TAB (Figure 5.1.D)

Using the center of rotation control on the center of rotation tab:

We have to opt one of the three choices that is the Automatic, Manual and Set value. It is

usually recommended to the user on using the ‘Automatic” as it is also the default

recommended method for the determination of the Center Of Rotation (COR). The “Manual”

method is only used whether the “Automatic” method fails. “Set value” method is used for

batches of identical components. We are referred on this study mainly in the “Automatic”

method and just briefly in the “Manual” method. It could be noticeable( ) that the “Automatic”

method is unable to estimate the center of rotation offset for a slice. This may be caused due

to poor contrast or one feature in the reconstructed image that may confuses the algorithm

estimating the COR location. If we change the height of the slice by a small amount and then

retry the algorithm will normally succeed.

However, if it fails we have to try the “Manual” method. When, the “Automatic” method fails,

an advanced feature enables the operator to visually appraise the sharpness of the image.

The “Manual” method works by reconstructing the a series of slices with divergent center or

rotation offset values. The user manually selects the slice which seems to be sharpest. When

a close match of values is found, the values should be noted and then the “Automatic’ method

is used co complete the calculation.

Using 1.Slice selection on the center of rotation tab:

In a manual mode, only one slice can be processed per time. Therefore, we should decide

whether we wish to use a single slice to set both the top and bottom COR values if it is

considered that they have little or no shear or either the top or bottom value in the case of

being the rotation sage tilted.

Using 4. Reconstruct (range for selected slice) tab:

If we wish, we may reconstruct the chosen slice to certify its sharpness. Furthermore, we may

regulate the parameters; Resolution and Quality. An example of an image in which the COR

is calculated, seems in Figure 5.1.E.

Using 5. Results, choose and apply tab:

Concerning each reconstructed result, the used value is displayed and we may use the radio

button to view the result. For the sample, an automatic value canculation is made and a

recommendation of the best result is given. When we select either a projection or

recommendation, this value is displayed in the Value to use control. This value is probably

modified. Then, we Apply this result in order to accept it.

5.5 USING THE “SET UP” TAB

Once the Center of Rotation calculation has been performed, our data will be sharp. By using

“Set up” tab other Set up corrections can be performed (display of ‘set up” controls and

displayed image viewer seem by Figure 5.1.F, Figure 5.1.G).

Using Interpolation tab:

This function chooses the value selection method when the back projected beam intersects

the projection image in order to get the value from the projection.

• Selecting “None” option:

The nearest pixel is chosen. This is marginally faster from the computation aspect but thi

gives a lower quality result than that of “Interpolation” function. This is not recommended

• Selecting “Interpolate” option:

The interpolated value is taken as the pre-assigned proportions of the four pixel neighborhood

where the beam intersects the image.

• Selecting “Reconstruct both” option:

This reconstructs all the above referred in (5.5 USING THE “SET UP” TAB) options.

Using Beam Hardening tab:

BACKGROUND:

When we take a solid uniform rod and look at one of its CT-slice images (projections) we will

see that edge is slightly brighter than the middle due to “Beam Hardening” factor. Most of the

lower energy X-rays are blocked to cross a larger distance than few kilometers of the sample,

as well as the remaining X-ray spectrum is shifted to a more energetic, a more penetrating

level. As a result of this, the linear X-ray attenuation displayed in the CT slices appears lower.

By analyzing the drop-off in X-ray attenuation in a material as a function of thickness

concerning the X-ray conditions in use, we can compensate for this during reconstruction. If

the sample is made of only one material, then this effect can be compensated for in software.

Otherwise, if the sample is made of more than one material, any result will be a compromise.

It is noteworthy being said that beam hardening correction is not performed by default. Beam

hardening is necessary for the case of hard materials such as steel whose thickness is larger

than this of few millimeters.

There are the two following sorts of Beam Hardening selection.

 Presets: This is a selection of typical values being used for increasingly hard materials.

We have to select “Reconstruction all presets’ to reconstruct all these preset and then choose

the value which most closely eliminates the effects of Beam Hardening.

 Custom: This is an advanced setting that requires the measurement of the flux density

through different thicknesses of sample material in order to make corrections that concern

Beam Hardening. (By desiring to use this selection, it needs for us to communicate with

‘Nikon Metrology” company by asking for the corresponding documentation).

USING SCATTER REDUCTION CONTROL:

This function is used to remove the excess intensity in the image which is caused due to

scattering. Thus, a value that removes half of the scattering could be doubled to remove all

the scattering. The image will seem to dim a little as this value is subtracted from the data.

Using Noise reduction control:

CT volumes are reconstructed by using a filtered back-projection method. The ideal

recommended filter being used in the absence of noise is a ramp filter. However, the existing

noise in the projection images may be enhanced by such a filter and other filters that may be

used in a compromise in order to reduce the resultant noise being presented in the CT

volumes.

A CT volume is formed by back-projecting a line from a pixel in a filtered projection image

through the volume back to the source. As each projection is added, the new line of

information is summed into the volume. The lines intersect in a region and due to this their

values reinforce each other. Where they do not intersect, they do not reinforce each other and

without the presence of a filter a highly background noise level would be generated. It is

noteworthy stating that the use of a correct filter causes this background value be cancelled

out to almost zero. When all the projections have been added, the volume is normalized to

the output range and the noise is correspondingly reduced to a quite low level. By taking more

projections, the signal-to-noise ratio is increased.

The noise can be removed in the slice view images by ignoring all values below a threshold

and only scaling the data above this.

The filtering choices being provided give a sliding scale between preserving detail but also

noise or removing noise but reducing detail.

Using Start Angle control:

This function allows the reconstruction being rotated based on the angle value being entered,

so that the reconstruction is orientated in a specific way. Also, this function can reduce the

volume size by making the object fit the region of interest on a better way.

5.6 USING THE “VOLUME” TAB

By using the “Volume” tab, we define the region of interest to be reconstructed that consists

the output volume (Figure 5.1.K).

The controls being included in the ‘Volume” tab (Figure 5.1.H) are described below.

Selection control:

The projections of the volume viewed by 0o and 90o are shown in Figure 5.1.I. By outlining the

reion of interest in both of these projection, this will ascertain that all parts contained within

this region will be reconstructed. By clicking on the thumbnail image we Can have the

projection viewed n the thumbnail image displayed on the large image window (image

displayer),as it seems on Figure 5.1.K

Radius of reconstruction control:

This function shows a top-down view of the thumbail image but as it is viewed from above. As

the selection range changes in “Selection above”, a rectangle is drawn, in the third dimension.

The radius of reconstruction corresponds to the width of the projection image as it is back

projected to the location of the object. However, some factors that are described below, can

reduce this width.

 Center of Rotation: The shift of the center of rotation each time reduces the

apparent width.

 Shuttling: This procedure is used in stop-start of the micro-CT scanning analysis to reduce

ring artifacts. However, due to this procedure, 10% of the projection width is lost.

The above functions can reduce the real radius of reconstruction by their application on a

combined way and if the Auto checkbox is selected, the calculation of this reduction can be

performed.

Resources control:

This control indicates the output size of the volume, the type of the computer that we need to

view the result on and the required memory capacity.

Resolution control:

This control permits to the user to specify the resolution of the generates voxels

corresponding to the volume. We normally have to choose100%. However, one user may

select values between 10% and 1000%. This is not recommended at it gives meaningless

results. It is always noteworthy holding in the x, y,z resolutions linked.

Volume control:

This volume control displays the numeric representation of the selected cuboiv volume.

Quality control:

This control displayes the percentage of used projections for the reconstruction data. This

always have to be 100%. This value have to be smaller when we want to get a quick low

resolution look at the voume aiming to ckeck the selected region.

Reconstruction control:

The reconstruction of the volumes is handled within this application.

Name control:

The name of the volume which will be created in the same directory as the current project is

defined on this control. The current parameter file will also be copid and given the same name

so that a record exists that shows the generation of the volume referred from the original

image data set. By default, the current name of the opened project is chosen and extended

with a numeric post-fix (i.e. if we open 01_Kat.xtekct.file then the output files will be

01_Kat_01.xtekct and 01_Kat_01.vgi file for the reconstruction, as it seems on (Figure

5.1.H)).

Start control:

This control submits the volume for reconstruction with the name displayed in “Name’ above.

If the reconstruction process is acceptable through the reconstruction engine, an “accepted”

message will be displayed (Figure 5.1.H) and “Name” will be updated to the next automatic

default name in the sequence.

5.7 USING THE “CALIBRATION” TAB

Calibration is only used if we wish to scale the output volume data so that specific areas are

characterized by a known output value. This procedure is used for scaling bone samples and

tissue to the Hounsfield Unit scale. The “Calibration” process requires the calibration of a

sample of the material that is usually provided as a cylinder of single material with a known

diameter. A radiograph of the sample material is taken under the same conditions as a CT-

scan of other samples containing the sample material. The calibration process is performed

through the followingly described controls.

Selection control:

This control is used to select the middle of the sample at its thickest position whether a

Computer Tomography (CT) scan of our calibration sample has been done.

Scaling control:

The normal scaling of the reconstruction volume is inverse meters (m-1).

The values of “Calibration Sample” and “Custom” are used with “Calibration sample scaling”.

Before the “Calibration Sample scaling” we can select “Calibration Sample, we need to do

“Set up” and evaluate the “Calibration Sample’ paramteters on the following way.

Calibration sample:

Calibration file control:

We browse and open the radiograph being taken from the calibration sample. If we have done

a CT-scan of the sample the browsed file will correspond to the first projection image of the

data set.

Area selection control:

The “red’ rectangle is placed over the image so that its densest part is centered in the middle

of the rectangle.

Material control:

The relative density of the material is chosen from the drop down list or “Custom” is selected

where the appropriate value has to be entered.

Sample diameter control:

The diameter of the material is defined through this control. If the material is a bottle of water,

this diameter value coincides with the inside diameter one.

Reconstruction control:

If we wish we may reconstruct the chosen slice to validate its sharpness. The

“Reconstruction” and “Quality” can also be regulated. When the reconstruction processs is

completed, we click on the reconstruction and pull the drawn square so that this covers the

ceneter of the calibration sample. Then, an average value will be displayed. It is noteworthy

being pointed out that if it is problematic for us to get the correct scaling factor, we may wish

to enter a “Custom” attenuation in “Scaling selection’ box being found on “Selection” control.

We get the correct scaling factor

by taking the Attenuation value that has been found and have this multiplied with the

percentage error of getting a “Custom” Attenuation value.

5.II VGLSTUDIOMAX SOFTWARE

VGLSTUDIOMAX software is useful as it reads the reonstructed (*.vgi files) that are created

by using CT-PRO software. In general, VGStudiomax is a useful software package for the

visualization and analysis of voxel data. Also, VGStudiomax software is used to process the

reconstructed projections of the dataset and help us calculate main properties characterizing

the structure of the bone.

5.II.A PREPARATIONS (WARNINGS) DURING VGSTUDIOMAX USE

The loading and processing of data in VGStudiomax might change the data without providing

the option of recovering their original state. Before using the software corresponded to our

own data, the creating of a backup copy is recommended in order to avoid unintended

changes.

Our working with voxel data demands considerable system resources. It would be better for

us to have all other applications shut down whether large amounts of memory or processor

power is consumed due to them.

In the case of being the system breakdown or being our system overcharged we might not be

able to to save or recover our work. Thus, it is highly recommended the saving of our work

regularly and especially after time-consuming or elaborating processes.

5.II.B WORKSPACE

The Workspace of VGStudiomax is divided into several workspace windows that are used by

a user to inspect and manipulate the objects that interest him. Resizing of the different

workspace windows is achieved by moving the cursor over their respective window borders.

The cursor is expected to be changed to a resize cursor indicating by its shape the direction

along which we can drag the cursor to resize the workspace window. By clicking inside a

workspace window, this will be marked as a currently selected workspace window. Each

workspace window is decorated with a set of controls allowing the user to modify the content

of a workspace window.

The user can observe a 3-D window displayed on VGStudiomax panel (Figure 5.2C). This

shows the result image of the rendering process.

The coordinate system tripod placed in the lower center corner of the 3-D window indicates

the orientation of the currently chosen coordinate system.

OBJECT SELECTION FROM THE SCENE TREE

When we select an object from the Scene Tree (and it is chosen to be visible by activating the

checkbox in front of it )(Figure 5.2C), the corner of a box surrounding this sill be shown in the

3-D window. Most of these corners will be “bright red” whether we are in the “Rotation” mode

and “bright green “ in the case of the ‘Move” mode.

5.II.C WORKSPACE WINDOW CONTROLS

Some of the most commonly useful “Workspace window controls” that are included in the

Table 5.2.A are described below.

Move/Rotate mode:

Two basic modes exist for the 3-D window which are the “Rotate” mode and the “Move”

mode. We can choose each one of the modes by clicking the respective icon in the icon bar,

by choosing the respective entry from the “Object menu” or by using the <Ctrl>+<Shift>+R /

<Ctrl>+<Shift>+M shortcuts.()

4/2/1 pixels per ray:

We have to specify if we want to render every 16th, every 4th or every pixel. The former option

leads to the fastest rendering, the latter to the most accurate.

Show scale bar:

This control toggles the display of the scale bar.

Show tripod:

This control toggles the display of the coordinate system tripod.

Show box:

Ths control toggles the display of the grid box of the coordinate system with tick marks.

Default camera views:

The corresponded icons provide shortcut to fourteen default camera views, relative to the

selected coordinates system. They can be used for quick viewing and generation of reference

screen shots. These default views are along the three major axes as well as from positions

along each of the four diagonals (one position for each corner).

Center and focus camera:

We click this button to center the currently selected object and make it become fully visible

within the 3-D window.

Freeze rendering:

We click this button to stop rendering of our object. This can be useful whether we want to

make changes to our data without being the results displayed in the 3-D window after each

step.

Toggle fullscreen state:

This control toggles the fullscreen mode of this window.()

Focus selected object:

This centers the slice view of the selected object in the 2-D window. By double-clicking

quickly we can first center the current slice and then center the data set.

Lock slice position:

This control locks the slice position for the 2-D window. This means that for example moving

the “Navigation cursor”( Figure 5.2.D) will not causes this window to switch to the respected

slice. However, it is possible to modify the slice position deliberately by entering a value in

slice position spin box

Toggle transform mode:

When we activate this button, we ca use the 2-D window to ”Rotate” or ”Move” an unlocked

object with respect to the currently chosen coordinate system.

When we are in the ”Move” mode, we can drag the object using the left mouse button. We

can hold down the middle mouse button over the object by and move the mouse up or down

to move the object forward or backward.

When we are in the ‘Rotate” mode, we can left-click the selected object, hold it and move the

mouse to rotate the object around the red cross displayed in this mode. We can hold down

the middle mouse button over the object and move the mouse up/down to rotate the object

clockwise/counter-clockwise.

Original display mode:

This toggles between displaying the slice images in their original brightness, contrast and

color settings and showing them in the settings specified in “Display” mode.

Color mode/ Color and Opacity mode:

This chooses which of the user-defined “Volume Rendering” (Figure 5.2C) settings (only the

colors and the opacity settings) we want being displayed on 2-D Windows.

Navigation plane:

This control toggles the display of the navigation planes in 2-D and 3-D windows. Navigation

planes indicate the position of the currently selected 2-D views in the data set. By grabbing

and dragging a navigation plane in 2-D or 3-D Window we can adjust the position of the

navigation plane.

2-D WINDOWS

The 2-D Windows display slices of the currently visible objects are they are seen from the

Front (along the y-axis), Top (along the z-axis), respectively Right (along the x-axis) of the

current coordinate system. There are three views which are active by default. Using the

Layout editor we can also add “Back”, “Bottom”, “Left” and “Rotation” views. Using the layout

editor we can also add “Back”, “Bottom”, “Left” and “Rotation” views. Voxel data are shown in

grey values. We can scroll through slices in a certain direction by scrolling the mouse wheel

while the mouse pointer is in the respective window or by changing the value in the text box,

in the lower left corner of the 2-D window.

5.II.D FILE MENU OPTIONS

The file menu of VGStudiomax software provides basic features for loading and saving

projects as well as for importing and exporting objects. The “File menu” allows us to handle

two basic file types which are object files and project files (or folders).

 Object files:

These files contain voxel-data representing user’s real object, basically the output of the CT-

scanner (after scanning and reconstruction) or some other device.

The number of these files per object varies. We may have one file contaning the whole object

(volume file formats) or an image stack (i.e. bitmap or tiff files) where each file represents one

slice of the scanned object .

 Project files (and folders):

*.vgi files contain basic information concerning the object including references to the object

files and supplementary files belonging to the project.

A *.vgi file must be accompanied by the project folder and the file(s) containing the object(s).

We can not a *.vgi file without the associated project folder and we can not load an object into

the scene without the files containing the object data.

Some of the most commonly useful “File menu optiions” that are included in the Table 5.2.B

are described below.

New:

We choose this option whether we want to exit the current project and create a new empty

scene.

Open:

This choice opens a project saved as a *.vgi. file. If we open a *.vgi (or *.vgl) file we have to

be sure that the project directory and object files are available as well.

Two examples of opened *.vgi. files are seen on Figure 5.2A., Figure 5.2.B.

Save:

This option saves the current project under the same *.vgl.filename and in the associated

project folder.

Merge object:

This choice merges objects derived from another project into the current scene. We have to

select the project which we want to import the object(s) from. We can import all the objects

from the other project or only a selection of them. In the “Merge object” dialog box, we select

the project needs being imported.

Import:

This option imports an object into the current project. The object can be voxel data or a

polygon object. We can import several objects into one scene.

Import an histogram:

We click the “Histogram button” in the ‘Load as’ dialog to open another window. The

“Histogram” tab displays the number of occurrences of each grey value in the object as a

diagram. The grey values of the voxel corresponded to the object are given on the axis of

ordinates. The dark grey area in the background of the diagram shows the number of

occurrences of each grey value.

The Histogram contains two vertical red lines, initially positioned at the very left and the very

right of the diagram. Each red line is labeled with two numbers, the first indicating the grey

value at the current position of the line, the second one shows how many voxels of the object

are of this grey value. We can move the red lines by left-clicking them. We shift them left or

right while holding of the left mouse button is pressed. Shifting a line to a new position will

adapt the value in the “Data range mapping” area of the “Load as” dialog and reversibly.

The “Calibration tab” permits us to perform a grey value calibration within the import process.

By using the “Calibration tab” we can select material and background grey values and assign

them to grey values considered as destination. We can select either the “Define background”

button or the “Define material” button. We can move the move the mouse cursor over the

preview window. We have to hold down the left mouse button and move the mouse to create

a selection rectangle. The area inside the rectangular will be considered for either background

or selection of the material.

Export:

This choice exports a single object, parts of an object or multiple objects of the current scene

as data files. The object(s) will be converted into the selected format.

5.II.E OBJECT MENU OPTIONS

The Object menu of VGStudiomax permits us to manipulate the objects in the scene (such as

voxel objects or polygon models, light sources, clipping planes).

Some of the mainly commonly useful “Object menu optiions” that are included in the Table

5.2.C are described below.

Clipping:

This option creates a clipping plane, a clipping box, an aligned clipping box or a clipping

polyline 3D for the selected object(s).

Surface determination:

The surface determination defines the material boundary. This boundary is expected being

displayed as a white line in the 2-D windows. We can use an automatic surface determination

or a determination based on sample area.

Standard surface determination)

The result of the “Define material by example area” option of the “Surface determination” is a

materal defined by one grey value applied globally to the object and delineated by a bright

surface line. This one grey value serves as a threshold. Brighter areas are considered as

material, darker areas are considered as background. The surface determination permits a

calculation of the material boundary in sub-voxel accuracy by tri-linear interpolation.

The presented table in the upper part of the dialog provides statistical information related to

the surgace determination. The “Status is undefined” until the surface determination is

performed. Depending on the surface determination mode, the “Status” is set to either

“Sample area” or “Automatic”. In case of a surface determination based on sample areas,

statistical information about the voxels covered by the area is provided that includes “Mean”

grey value, “Standard deviation” of the grey values, “Minimum” and “Maximum” grey value

within the respective areas.

The histogram in the middle of the dialog displays the grey value distribution of the voxels in

the scanned volume. The left-most peak usually represents the background voxels. The zoom

of the y-axis can be modified by clicking in the histogram and dragging the mouse cursor

downwards for full view of large peaks or upwards for identification of minute peaks.

We can control the surface determination through the histogram as it is described below.

The histogram contains a red “isorface” line indicating the grey value which the yellow surface

line in he 2-D Windows is calculated for. We use the mouse to shift this line and then the slice

views are updated accordingly. Shifting the line on the right will fit the isosurface tighter to the

object.

After a surface determination, two blue lines will be positioned in the peaks of what has been

defined or detected as background and material. The mouse is used to shift these lines. The

“red” isosurface line will be shifted such that it is always exactly in the middle of the two blue

lines and then the slice views will be updated accordingly.(Figure 5.2.E)

 Specify sample area for surface determination:

We click the “Define material by example area” button to perform a manual surface

determination. We will be prompted to draw a typical background area with the mouse

touched in a slice window. A selection will be created on this way. The standard selection

mode for the surface determination is the “Rectangle”. On a next stage, we click “Next” to

proceed with a similar selection of a typical material area. (Figure 5.2F) Then, we click

“Finish” to have actually the surface determination performed, based on our selections.

 Automatic surface determination:

We click the ‘Automatic” button to perform a surface determination based on the

histogram.(Figure 5.2.E) VGLStudiomax software determines the background peak and the

material peak in the histogram and then this calculates the grey value of the material

boundary. Automatic surface determination is recommended for the case of volumes being

consisted of objects corresponding to one material exclusively.

Remove Surface determination:

This option deletes the surface determination information. After removing the surface

determination information, the white line delineating the material boundary is no longer visible

and the material boundary is no longer defined.

Properties:

This choice displays information about the selected volume object.

General (Volume objects) dialog:

The “Object name’ of the selected object is editable. The overall scanned area is quantified in

the “Data set/Region of interest info”. It is noteworthy being said that the overall scanned area

usually comprises the surrounding air, not only the scanned object. The bounding box is

displayed in 3-D window.

The bounding box related information contains the “Dimensions [voxel]”, the Resolution’ on

the x, y and z axis, the “overall voxel count” in the bounding box, the “Total volume” and the

“Dimensions”.

The “Object info’ provides information on the scanned object. If there is no available surface

determination, the information is calculated by using the manually set “Isosurface value”. The

object information contains the “Volume’ of the object, the remaining ‘Total volume-object

volume”, the “Surface area” of the object, the “Closed surface area” including the material

surface as well as the ROI surface inside of the material, the “Surface area difference’

between the “Closed surface area’ and the “Surface area and the dimensions of the smallest

possible bounding box around the object as well as clicking “Update” to re-calculate the

parameters. (VGStudiomax software manual)

Morphometrics dialog:

This shows information concerning morphometrical indices. Their calculation depends on the

material (bone) surface (BS), the material (bone) volume (BV) and the total volume (TV). The

morphometrical indices are described as follows(VGStudiomax software manual).

 BS/TV: the ratio of the material (bone) volume to total volume

 BS/BV: the ratio of material (bone) surface (BS) to material (bone) volume (BV)

 Tb.Th: the mean thickness of trabecular structures in the material (bone)

 Tb.N: the mean number of trabecular structures per unit length

 Tb.Sp: the mean distance between trabecular structures

5.II.F SELECT MENU OPTIONS

The “Select’ menu allows us to create 3-D region masks from an object as well as enabling

the manipulation of the region masks. A region is a mask covering part(s) of an object which

is of special interest (region of interest (ROI)). A ROI can be used to confine the calculation of

any analysis function within the volume covered by the ROI.

The ‘Selection modes” of VGStudiomax software are used to create and modify ROI masks.

First, a temporary selection is made by using one of the “Selection modes”. Then a ROI is

created directly from the temporary selection or it is added or subtracted from the temporary

selection from or to the currently selected ROI. The contour of the “selection” can be created

in any of the available 2-D views. The “Selection modes” only modify the ROI mask without

modifying the actual volume data.

Some of the mainly commonly useful “Select menu optiions” that are included in the (Table

5.2.D, Table 5.2.D(supplement)) are described below.

Draw:

This mode paints directly into the selection mask by using a spherical or cylindrical

brush.(.VGStudiomax manual.)

Rectangle:

This mode creates a rectangular selection in the chosen 2-D view. We press and hold the left

mouse button to define the first corner of the rectangular, drag across the 2-D view and

release the mouse button again to set the opposite corner.

By pressing and holding the left mouse button over any of the red corners, the rectangular

selection can be resized by dragging the corner into a new position. We have to press and

hold the left mouse button over the yellow rectangle and drag in order to reposition this.

In order to extend our selection into the third dimension we have to press and hold the left

mouse button over the red top/button markers in any of the two other 2-D views. We can

extend our selection by dragging the marker into the desired position.

Region growing:

This “Selection menu’ option creates a selection by using a region growing algorithm. After

selecting an initial seed pointing any of the three 2-D views, the algorithm will expand the

“selection”. The expansion continues as long as new voxels are found which are connected to

the growing selection and where the data value is within the specific tolerance relative to the

initial seed point or growth criterion.

We can select the seed point for the “region growing” by pressing the left mouse button over

the desired location in any of the three 2-D Windows. In order to limit the extent of the region

growing within a sphere, we have to keep the button pressed and drag so that we define the

radius of the sphere. Once the mouse button is released, the region growing will automatically

begin. Changing any of the region growing parameters will restart the process by using the

new values/settings and the last defined seed point.

Adaptive rectangle:

This “Selection mode” is consistent to an iterative procedure of finding the edges of an object

inside a rectangular search region by using the edges of the rectangular region as an initial

guess. Initially, the full rectangular volume is selected. Then after each iterative step,

background material is removed.

We have to press and hold the left mouse button in any of the three 2-D windows to set the

first corner of the rectangular area. We have to drag and release in order to set the opposite

corner. The rectangular region can be resized by dragging any of the corners into a new

position. In order to extend the selection into the third dimension we have to press and hold

the left mouse button over the red top/button markers in any of the two other 2-D views. The

selection can be extended by dragging the marker into the desired position.

5.II.G ANALYSIS MENU OPTIONS

We can have an access on tools providing advanced analysis features through “Analysis”

menu.

Some of the mainly commonly useful “Select menu optiions” that are included in the Table

5.2.E, are described below.

Wall thickness:

This option calculates and visualizes the material thickness of a voxel object.

Defect detection:

This option calculates and visualizes the material defects of a voxel object.

The defect detection procedure consists of two steps (VGStudiomax software manual):

 Each voxel is checked whether it may be a part of a defect or not. This step creates group

consisted of connected defect candidates.

 Each group of defect candidates is checked whether it fits the parameters specified by the

user.

The “Default” algorithm that always has to be selected by the user is allowed to be used for

grey value variations by detecting defects (i.e. dark areas).

New analysis annotation:

This option creates an annotation attached to the selected analysis. An annotation can be

used to show analysis results and data values at a specific location in the dataset.

Volume analyzer:

This tool (Analysis>Volume Analyzer) is utilized to calculate the various properties of the

voxels in the selected volume or Region of Interest (ROI).

Data histogram:

This histogram analyzes the voxel data within the selected region. The histogram area shows

the grey value distribution/histogram of the data within the selected region. The grey value

range of the voxels to be analyzed can be specified by moving the two red sliders, the left and

the right slider setting the minimum and maximum grey value respectively. Each slider

indicates the grey value at the slider position, as well as the total number of voxels having the

specific grey value.

When the “Values’ checkbox is on the following properties of the voxels in the selected region

and grey values range are calculated the “Min”, “Max” show the minimum/maximum gray

value of the voxels within the selected region and gray value range, the “Mean” and

“Deviation” of the voxels within the selected region and gray value range and the “Volume”

and “Number of voxels” the geometric volume and the total number of voxels within the

selected region and gray value range.

Cursor range:

The “cursor range” shows the minimum and maximum grey value range of the interval that

has to be analyzed.

Histogram range:

The Histogram ranges regulates the grey value range of the primary axis of the histogram.

5.II.G CHARACTERIZATION OF THE QUALITY OF RECONSTRUCTED IMAGES

The quality of high-resolution micro-CT images is often corrupted by ring artefacts. These

concentric ring artifacts are partially seen in our reconstructed images (Figure 5.2A, Figure

5.2B). These artifacts are of white color and they are occasionally discerned near the pores

being presented on the bone surface. Also, this sort of discrepancies are noticed in the

reconstructed image as concentric-ring like artifacts that are superimposed onto the initial

image which is produced by the micro-CT scanning analysis. The appearance of these

discrepancies can possibly caused due to imperfectness related to the performance of the

detector elements that participate in the micro-CT scanning of the bone samples. By

Kyriakou, imperfect of defect detector elements may cause concentric-ring artifacts due to

their continuous over- or underestimation of attetunation values. Indeed, the appearance of

the ring artifacts could be possibly affected by the movement of the stack of the specimens.

Furthermore, it could be considered that the existence of these discrepancies could occur due

to the high number of rotations which take place inside the micro-CT scanning system.

The consequences of the presence of ring atifacts on the bone surface concern impairing of

the quality of micro-CT reconstructed images. The existence of ring artifacts can possibly

affect the correct calculation of the center of rotation in the reconstructed images. Due to the

appearance of concentric ring artifacts, the function of the beam hardening and the noise

reuction filters is hampered. As an effect of this, an over-estimation or under-estimation of the

physical properties of bone specimens occurs. Moreover, the appearance of ring artifacts

impairs the quality of the medical imaging.

The ring artifacts could be erased from the reconstructed images by the implementation of

several smoothing filters to the reconstructed images. A possible establishment of an

algorithm which would effectively identify the ring artifacts or/and implement a type of

correction on the processing of reconstructed images could improve the quality of

reconstructed images by disappearing or reducing the appearance of concentric ring artifacts

on the scan-imaging.

6. GENERAL DISCUSSION

The mechanical properties of bones and their loading conditions area a main determinant of

the risk of bone fractures.(118) This situation is dominantly presented on elderly people due to

the risk of osteoporosis on them but also the greatly low rate of bone regeneration. The

mechanical response of bone is mainly determined by its density and characteristics of its

structure. The estimation of mechanical properties of trabecular bone regions is of particular

significance because the osteoporotic fractures initially appear in these regions.Apart from

destructive mesurements such as experimental assessments, non-destructive thechniques

are currently used for the determination of bone density.(118) and architectural parameters of

bone tissue

Among the non-destructive techniques of evaluating architecture of bone, X-ray micro-

computative tomography (micro-CT) has been widely used for the characterization of the

trabecular bone structure. (6) This constitutes an advantageous technique for the assessment

of trabecular bone(104) tissue due to its being a non-destructive method, avoids specimen

preparation and provides three-dimensional (3-D) images with a high and isotropic spatial

resolution up to a few micro-meters in the three spatial directions(153).

There have been studies(118) which search for the assessment of bone structure parameters

and the assignment of mechanical properties of bone structures by calculating them using

Finite Element (FE) techniques applied on scanning images (images of bone specimens

obtained by using micro-CT scanner). Micro-CT imaging techiques allow the generation of

micromechanical models of the scanned tissue by directly converting bone voxels to

mechanical elements that can be analyzed by the finite element (FE) method. Moreover, the

incorporation of micro-CT information into

(micro)-mechanical models can consist a powerful tool for the assessment of bone

mechanical strength and the quantification of tissue stress and strain distributions under

physiologic and applied loading conditions. It would be important to realize that only the

spatial integration of physical material properties into these models will give a true estimate of

bone quality.

REFERENCES

1. Y. H. An, W. R. Barfield, R. A. Draughn (2000) Mechanical Testing of Bone and the Bone-Implant

Interface. Chapter 2: Basic Concepts of Mechanical Property Measurement and Bone Biomechanics.

Edited by Y.H. An and R.A. Draughn. CRC Press LLC, p. 23-39.

2. M. J. Anderson, J. H. Keyak, H. B. Skinner (1992). Compressive mechanical properties of human

cancellous bone after gamma irradiation. J. Bone Joint Surg.; Vol.74-A, p. 747-752.

3. R. B. Ashman, J. Y. Rho, C. H. Turner (1989). Anatomical variation of orthotropic elastic moduli of

the proximal human tibia. J. Biomech.; Vol. 22, p. 895-900.

4. R. M. Aspden, B. Li (1998). Reproducibility using Archimedes’ principle in measuring cancellous

bone tissue by L. Zou, R. D. Bloehaum and K. N. Bachus. Medical Engineering and Physics; Vol. 20 p.

159-160.

5. X. Banse, J. P. Devogelaer, A. Lafosse et al. (2002a) Cross-link Profile of Bone

Collagen Correlates with Structural Organization of Trabeculae. Bone; Vol. 31(1), p.70-76.

6. J. F. Barrett, N. Keat (2004). Artifacts in CT; Recognition and Avoidance. Radiographics; Vol. 24, p.

1679-1691.

7. A. Basillais, S. Bensamoun, C. Chappard, B. Brunet-Imbault, G. Lemineur, B. Ilharreborde, M.-C.

Ho Ba Tho, C.-L. Benhamou (2007). Three-dimensional characterization of cortical bone microstructure

by microcomputed tomography: validation with ultrasonic and microscopic measurements. J. Orthop.

Sci.; Vol. 12, p. 141-148.

8. E. Bonucci (2000). Mechanical Testing of Bone and the Bone-Implant Interface. Chapter 1: Basic

Composition and Structure of Bone. Edited by Y.H. An and R.A. Draughn. CRC Press LLC, p. 3-17

9. P-L. Bossa r t , H. E. M a r t z, H. R. Brand, K. Hollerbach (1995). Application of 3D X-Ray CTData

Sets to Finite Element Analysis. This paper was prepared for submittal to the 22 Quantitative Non-

destructive Evaluation Conference Seattle, WA. www.osti.gov/bridge/servlets/purl/119981-

MNaDa/webviewable/

10. SK. Boyd, R. Muller, RF Zernicke (2002). Mechanical and architectural bone adaptation in early

stage experimental osteoarthritis. J. Bone Miner. Res.; Vol. 17, p. 687-694.

11. M. D. Brodt, G. B. Pelz, J. Taniguchi, M. J. Silva (2003). Accuracy of peripheral quantitative

computed tomography (pQCT) for assessing area and density of mouse cortical bone. Calcif. Tissue

Int.; Vol. 73, p. 411-418.

12. J. A. Buckwalter, M. J. Glimcher, R. R. Cooper, R. Recker (1995) Bone Biology, Part I and Part II.

The Journal of Bone and Joint Surgery; Vol. 77-A, No. 8, p. 1255-1290.

13. W.T. Butler (1984). Matrix Molecules of Bone and Dentin. Collagen Related Research; Vol. 4,

p.297.

14. J. Canny (1986). A Computational Approach to Edge Detection. IEEE Transactions on Pattern

Analysis and Medicine Intelligence; Vol. 8(6), p. 679-697.

15. D. R. Carter, W. C. Hayes (1977). The compressive behavior of bone as a two-phase porous

structure,. J. Bone Joint Surg. Am.; Vol. 59, p. 954.

16. D. R. Carter, W. C. Hayes (1977). The compressive behaviour of bone as a two-face porous

structure. J. Bone Jt Surg.; Vol. 59A, p. 954-962.

17. A. Chawla, S. Mukherjee, G. Sharma. Finite Element Meshing of Human Bones from MRI/CT Raw

Data.

18. E. A. Chrishilles, C. D. Butler, C. S. Davis, R. B. Wallace (1991). A model of lifetime osteoporosis

impact. Arch. Intern. Med.; Vol. 151, p. 2026-2032.

19. A. M. Coats, P. Zioupos, R. M. Aspden. (2003). Material properties of subchondral bone from

patients with osteoporosis or osteoarthritis by microindentation testing and electron probe microanalysis.

Calcified Tissue International Vol.; 73 p. 66-71.

20. D. D. Cody, G. J. Gross, F. J. Hou, H. J. Spencer, S. A. Goldstein, D. P. Fyhrie (1999). Femoral

strength is better predicted by finite elements than qct and dxa. J. Biomechanics; Vol. 32, p. 1013-1020.

21. P. G. Coelho, P. R. Fernandes, H. C. Rodrigues, J. B. Cardoso, J. M. Guedes (2009). Numerical

modelling of bone tissue adaptation-A hierarchical approach for bone apparent ,density and trabecular

structure. Journal of Biomechanics; Vol. 42, p. 830-837.

22. J. Cohen, W. H. Harris (1958). The three dimensional anatomy of the haversian system. J. Bone Jt.

Surg.; Vol. 40-A, p. 419.

23. R. B. Cook. (Academic year 2005-2006). Non-Invasively Assessed Skeletal Bone Status and its

Relationship to the Biomechanical Properties and Condition of Cancellous Bone. PhD Thesis.

24. D. Cooper, A. Turinsky, C. Sensen, B. Hallgrimsson (2007). Effect of voxel size on 3D Micro-CT

Analysis of Cortical Bone Porosity. Calcif. Tissue Int.; Vol. 80, p. 211-219.

25. C. Cooper, E. J. Athkinson, S.J. Jacobsen, W. M. O’ Fallon, L. J. Melton (1993). Population-based

study of survival after osteoporotic fractures. Am. J. Epidemiol.; Vol. 137, p. 1001-1005.

26. B. Couteau, Y. Payan, S. Lavallee, (2000). The mesh-matching algorithm: an automated 3d mesh

generatof for finite element structures. J. Biomechanics; Vol. 33, p. 1005-1009.

27. J. D. Currey (2002). Bones. Structure and Mechanics. Published by Princeton University Press;

Chapter 1: The structure of Bone Tissues, p. 22, 23.

28. J. D. Currey (1988). The effect of porosity and mineral content on the Young’s modulus of elasticity

of compact bone. J. Biomech.; Vol. 21, p. 131-139.

29. J. D. Currey (1988). Strain rate and mineral content in fracture models of bone. J. Orthop. Res.; Vol.

6, p. 32-38.

30. M. Ding, A. Odgaard, F. Linde, I. Hvid (2002). Age-related variations in the microstructure of human

tibial cancellous bone. J. Orthop. Res.; Vol. 20, p.615–621, doi:10.1016/S0736-0266(01)00132-2.

31. M.Ding, A. Odgaard, I. Hvid (1999). Accuracy of cancellous bone volume fraction measured by

micro-CT scanning. J. Biomechanics; Vol.32, p. 323-326.

32. T. Dufresne (1998). Segmentation techniques for analysis of bone by three-dimensional computed

tomographic imaging. Technol Health Care; Vol. 6, p. 351–359.

33. A. Elmoutaouakkil, F. Peyrin, J. Elkafi, AM Laval-Jeantet (2002). Segmentation of cancellous bone

from high-resolution computed tomography images: Influence on trabecular bone measurements. IEEE

Trans Med Imaging; Vol.21, p. 354–362.

34. N. L. Fazzalari, D. J. Crisp, B. Vernon-Roberts (1989). Mathematical modelling of trabecular bone

structure: the evaluation of analytical and quantified surface to volume relationships in the femoral head

and iliac crest. J. Biomech.; Vol. 22, p. 901-910.

35. N. L. Fazzalari, J. Darracott, B. Vernon-Roberts (1985). Histomorphometric changes in the

trabecular structure of a selected stress region in the femur in patients with osteoarthritis and fracture of

the femoral neck. Bone; Vol. 6, p. 125-133.

36. N. L. Fazzalari, J. Darracott, B. Vernon-Roberts (1983). A quantitative description of selected stress

regions of cancellous bone in the head of the femur using automatic image analysis. Metab. Bone. Dis.

Rel. Res.; Vol. 5, p. 119-125.

37. L. A. Fedcamp, S. A. Goldstein, A. M. Parfitt, G. Jesion, M. Klerekoper (1989). The direct

examination of three-dimensional bone architecture in vitro y computed tomography. J. Bone Miner.

Res.; Vol. 4, p. 3-11.

38. Finnilä, M. A.J., Zioupos, P., Herlin, M., Miettinen, H. M., Simanainen, U., Håkansson, H.,

Tuukkanen, J. (2010). Effects of 2,3,7,8-tetrachlorodibenzo-p-dioxin exposure on bone. Journal of

Biomechanics; Vol.43, p. 1097-1103.

39. D. P. Fyhrie, S. M. Lang, S. J. Hoshaw, M. B. Schaffler, R. F. Kuo (1995). Human vertebral

cancellous bone surface cdistribution. Bone; Vol. 17(3), p. 287-291.

40. D. P. Fyhrie, N. L. Fazzalari, R. Goulet, S. A. Goldstein (1993). Direct calculation of the surface-to-

volume ratio for human cancellous bone. J. Biomechanics; Vol. 26(8), p. 955-967.

41. J. Galante, W. Rostoker, R. D. Ray (1970). Physical properties of Trabecular bone. Calcified Tissue

Research; Vol.5 p. 236-246.

42. I. J. Gibson (1985). The mechanical behaviour of cancellous bone. J. Biomech.; Vol. 18; p. 317-

328.

43. L. J. Gibson, M. F. Ashby (1997). Cellular solid: Structure and Properties. Cambridge University

Press, Cambridge, UK.

44. S. Goldstein (1987). The mechanical properties of trabecular bone: dependence on anatomic

location and function. J. Biomech.; Vol. 20 p. 1055-1061.

45. B. R. Gomberg, P. K. Saha, H. K. Song, S. N. Hwang, F. W. Wehrli (2000). Topological Analysis of

Trabecular Bone MR Images. IEEE Trans Medical Imaging;, Vol. 19(3), p. 166 -174.

46. M. Gresle (2002). Microstructural heterogeneity and fatigue life of human bone. Report of an

internship project.

47. T. Hara, E. Tanck, J. Homminga, R. Huiskes (2002). The influence of microcomputed tomography

threshold variations on the assessment of structural and mechanical trabecular bone properties. Bone;

Vol. 31, p. 107–109. doi:10.1016/S8756-3282(02)00782-2.

48. C. J. Hernandez, G. S. Beaupré, T. S. Keller, D. R. Carter (2001). The influence of bone volume

fraction and ash fraction on bone strength and modulus. Bone; Vol. 29(1), p. 74-78.T

49. . Hildebrand, P. Ruegsegger (1997). A new method for the model -independent assessment of

thickness in three -dimensional images. J Microscopy; Vol.185, p. 67 -75.

50. J. H. Hubbel, S. M. Seltzer (1995). Tables of X-Ray Mass Attenuation Coefficients and Mass

Energy-Absorption Coefficients. NISTIR 5632. NIST. US Department of Commerce, Gaithersburgh, MD,

USA.

51. I. Hvid, S. M. Bentzen, F. Linde, I. Mosekilde, B. Pongsiopetch (1989). X-ray quantitative computed

tomography: The relations to physical properties of proximal tibial trabecular bone specimens. J.

Biomech.; Vol. 22, p. 837-844.

52. Y. Jiang, J. Zhao, E. Y. Liao, R. C. Dai, X. P. Wu, H. K. Genan(2005). Application of micro-CT

assessment of 3-D bone microstructure in preclinical and clinical studies. J. Bone Miner. Metab.; Vol.

23(Suppl), p.122–131.

53. L. C. Johnson (1966). The kinetics of skeletal remodelling. Birth Defects Original Article Series;

Vol.2, p. 66.

54. S. Judex, S. Boyd, Y-X Qin, L. Miller, R. Müller, C. Rubin (2003). Combining High-resolution Micro-

computed tomography with material composition to define the quality of Bone Tissue. Current

Osteoporosis Reports; Vol. I, p. 11-19.

55. P. Kalebo, K. G. Strid (1988). Bone mass determination from microradiographs by computer-

assisted videodensitometry. II. Aluminium as a reference substance. Acta Radiol.; Vol. 29, p. 611-617.

56. W. A. Kalender (1992). A phantom for standardization and quality control in spinal bone mineral

measurements by QCT and DXA: design considerations and specifications. Med. Phys.; Vol. 19, p. 583-

586.

57. T. S. Kaneko, J. S. Bell, M. R. Pejcic, J. Tehranzadeh, J. H. Keyak (2004). Mechanical properties,

density and quantitatitative CT scan data of trabecular bone with and without metastases. Journal of

Biomechanics; Vol. 37, p. 523-530.

58. M. Karolczak, S. Schaller, K. Engelke, A. Lutz, U. Taubenreuther, K. Wiesent, W. Kalender (2001).

Implementation of a cone-beam reconstruction algorithm for the

single-circle source orbit with embedded misalignment correction using homogeneous coordinates.

Med. Phys.; Vol. 28, p. 2050–2069,doi:10.1118/1.1406514.

59. M. Kasra, M. D. Grynpas (1994). Effect of long-term ovariectomy on bone mechanical properties in

young female cynomolgus monkeys. Bone; Vol.15, p. 557.

60. T. M. Keaveny, T. P. Pinilla, R. P. Crawford, et al. (1997) Systematic and random errors in

compression testing of trabecular bone. J. Orthop. Res.; Vol. 15, p. 101.

61. T. M. Keaveny, R. E. Borchers, L. J. Gibson, W. C. Hayes (1993). Trabecular bone modulus and

strength can depend on specimen geometry. J. Biomech.; Vol. 26, p. 991-1000.

62. T. M. Keaveny, R. E. Borchers, L. J. Gibson, W. C. Hayes (1993). Theoretical analysis of the

experimental artefact in trabecular bone compressive modulus. J. Biomech.; Vol. 26, p. 599-607.

63. M. J. Keenan, M. Hegsted, K. L. Jones, J. P. Delany, J. C. Kime, L. E. Melancon, R. T. Tulley, K. D.

Hong (1997). Comparison of bone density measurement techniques: DXA and Archimedes’ principle.

Journal of Bone and Mineral Research; Vol. 12, p. 1903-1907.

64. T. S. Keller (1994). Predicting the compressive mechanical behavior of bone. Journal of

Biomechanics; Vol. 27, p. 1159-1168.

65. J. H. Keyak, S. A. Rosi, K. A. Jones, H. B. Skinner (1998). Prediction of femoral fracture load using

automated finite element modelling. Journal of Biomechanics; Vol. 31, p. 125-133.

66. J. H. Keyak, I.Y. Lee, H. B. Skinner (1994). Correlations between orthogonal mechanical properties

and density of trabecular bone: Use of different densitometric measures. Journal of Biomedical Materials

Research; Vol. 28, p. 1329-1336.

67. D.-G. Kim, G. T. Christopherson, X. N. Dong, D. P. Fyhrie, Y. N. Yeni (2004). The effect of

microcomputed tomography scanning and reconstruction voxel size on the accuracy of stereological

measurements in human cancellous bone. Bone; Vol. 35, p. 1375-1382.

68. L. Knott, A. J. Bailey (1998) Collagen Cross-Links in Mineralizing Tissues: A

Review of Their Chemistry, Function and Clinical Relevance. Bone; Vol. 22(3), p.181-187

69. M. Kothari, T. M. Keaveny, J. C. Lin, D. C. Newitt, H. K. Genant, S. Majumdar (1998). Impact of

spatial resolution on the prediction of trabecular architecture parameters. Bone; Vol. 22, p. 437-443.

70. JL Kuhn, SA Goldstein, LA Feldkamp, RW Goulet, G. Jesion (1990). Evaluation of a

microcomputed tomography system to study trabecular bone structure. J Orthop Res; Vol. 8, p. 833–

842.

71. Y. Kyriakou, D. Prell, W. A. Kalender (2009). Ring artifact correction forvhigh-resolution micro CT.

Phys. Med. Biol.; Vol. 54, p. N385-N391.

72. A. Laib, J. L. Kumer, S. Majumdar, N. E. Lane (2001). The temporal changes of trabecular

architecture in ovariectomized rats assessed by MicroCT. Osteoporos. Int.; Vol. 12, p. 936–941,.

doi:10.1007/s001980170022.

73. A. Laib, Ruegsegger P (1999). Comparison of structure extraction methods for in vivo trabecular

bone measurements. Comput Med Imaging Graph; Vol. 23, p. 69–74.

74. W. J. Landis (1995) The strength of a calcified tissue depends in part on the molecular structure

and organization of its constituent mineral crystals in their organic matrix. Bone; Vol. 16, p. 533-544.

75. C. M. Langton, T. J. Haire, P. S. Ganney, C. A. Dobson, M. J. Fagan (1998). Dynamic stochastic

simulation of cancellous bone resorption. Bone, Vol. 22, p. 375 -380.

76. C. M. Les, J. H. Keyak, S. M. Stover, K. T. Taylor, A. J. Kaneps (1994). Estimation of material

properties in the equine metacarpus with use of quantitative computed tomography. Journal of

Orthopaedic Research; Vol. 12, p. 822-833.

77. F. Linde, I. Hvid, F. Madsen (1992). The effect of specimen geometry on the mechanical behaviour

of trabecular bone specimens. J. Biomech.; Vol. 25, p. 359-368.

78. F. Linde, B. Pongsoipetch, L. H. Frich, I. Hvid (1990). Three-axial strain controlled testing applied to

bone specimens from the proximal tibial epiphysis. J. Biomech.; Vol. 23, p. 1167-1172.

79. F. Linde, I. Hvid (1989). The effect of constraint on the mechanical behaviour of trabecular bone

specimens. J. Biomech.; Vol. 22, p.484-490.

80. F. Linde, I. Hvid, B. Pongsoipetch (1989). Energy absorptive properties of human trabecular bone

specimens during axial compression. J. Orthop. Res.; Vol. 7 p.432.

81. J. C. Lotz, E. J. Cheal, W. C. Hayes (1991). Fracture prediction for the proximal femur using finite

element models: part 1-linear analysis. J. of Biomechanical Eng.; Vol. 113, p. 353-360.

82. A. Lu, Y. Ries, S. Majumdar (2003). A micro -computed tomography study of the trabecular bone

structure in the femoral head,” J Musculoskel Neuron Interact; Vol.3(2), p. 176 -184..

83. J. Magaziner, E. M. Simonsick, T. M. Kashner, J. R. Hebel, J. E. Kenzora (1989). Survival

experience of aged hip fracture patients. Am. J. Public. Health; Vol. 79, p. 274-278.

84. M. Martens, R. Van Audekercke, P. Delport, et al. (1983). The mechanical characteristics of

cancellous bone at the upper femoral region. J. Biomech.; Vol. 16 p. 971.

85. B. B. Martin, D. R. Burr, N. A. Sharkey (1998) Skeletal Tissue Mechanics. Springer, New York.

86. R. B. Martin, R. M. Rose (1984) Porosity and specific surface of bone. CRC Critical Reviews in

Biomedical Engineering; Vol. 10, No.5 p. 179-225.

87. J. A. Meganck, K. M. Kozloff, M. M. Thornton, S. M. Broski, S. A. Goldstein (2009). Beam hardening

artifacts in micro-computed tomography scanning can be reduced by X-ray beam filtration and the

resulting images can be used to accurately measure BMD. Bone; Vol. 45, p. 1104-116.

88. S. S. Mehta. (1995) Analysis of the mechanical properties of bone material using non-destructive

ultrasound reflectometry. PhD Dissertation, The University of Texas Southwestern Medical Center at

Dallas.

89. B. Mertz, P. Niederer, R. Muller, P. Ruegsegger (1996). Automated finite element analysis of

excised human femura based on precision-qct. J. Biomechanical Eng.; Vol. 118, p. 387-390.

90. G. A. Mohr, F. Little (1994). Effective 3D Geometry Extraction and Reverse CAD Modeling. Review

of Progress in QNDE; Vol. 14, p. 651-656.

91. E. F. Morgan, H. H. Bayraktar, T. M. Keaveny (2003). Trabecular bone modulus-density

relationships depend on anatomic site. Journal of Biomechanics; Vol. 36, p. 897-904.

92. R. Muller, B. Koller, T. Hildebrand, A. Laib, S. Gianolini, P. Ruegsegger (1998). Resolution

dependency of microstructural properties of cancellous bone based on three-dimensional mu-

tomography. Technol. Health Care; Vol. 4, p. 113-119.

93. R. Muller, H. Van Campenhout, B. Van Damme, et al. (1998). Morphometric analysis of human

bone biopsies: a quantitative structural comparison of histological sections and microcomputed

tomography. Bone; Vol. 23, p. 59-66.

94. R. Müller, P. Rüegsegger (1997). Micro-tomographic imaging for the nondestructive evaluation of

trabecular bone architecture. Stud. Health Technol. Inform.; Vol. 40, p. 61-79.

95. R. Muller, T. Hildebrand, P. Ruegsegger (1994). Non-invasive bone biopsy: a new method to

analyse and display the three-dimensional structure of trabecular bone. Phys. Med. Biol.; Vol. 39, p.

145-164.

96. L. M. Namara, P. J. Prendergast, M. B. Schaffler (2005). Bone tissue material properties are altered

during osteoporosis. Journal of Musculoskeletal and Neuronal Interactions; Vol. 5, p.342-343.

97. P. H. F. Nicholson, X. G. Cheng, G. Lowet, S. Boonen, M. W. J. Davie, J. Dequeker, G. Van der

Perre (1997). Structural and material properties of human vertebral cancellous bone. Medical

Engineering and Physics; Vol. 19, p.729-737.

98. S. Nuzzo, M. H. Lafage-Proust, E. Martin-Badosa, G. Boivin, T. Thomas, C. Alexandre, F. Peyrin

(2002). Synchrotron radiation microtomography allows the analysis of three-dimensional

microarchitecture and degree of mineralizaion of human iliac crest biopsy specimens: effects of

etidronate treatment. J. Bone Miner. Res.; Vol.17, p. 1372-1382.

99. A. Odgaard, I. Hvid, F. Linde (1989). Compressive axial strain distributions in cancellous bone

specimens. J. Biomech; Vol. 22, p. 829.

100. A. Odgaard (1997). Three -dimensional methods for quantification of cancellous bone

architecture. Bone; Vol. 20(4), p. 315 -328.

101. A. Odgaard, I. Hvid, F. Linde (1989). Compressive axial strain distributions in cancellous bone

specimens. J. Biomech.; Vol. 22, p. 829-835.

102. O. B. Olurin, M. Arnold, C. Korner, R.F. Singer (2002). The investigation of morphometric

parameters of aluminum foams using micro -Computed tomography. MaterialsScience and Engineering;

Vol. A328, p. 334 -343.

103. H. Oxlund, L. Mosekilde, G. Ørtoft (1996). Reduced Concentration of Collagen Reducible Cross

Links in Human Trabecular Bone with Respect to Age and Osteoporosis. Bone; Vol. 19(5), p.479-484.

104. M. Parfitt,, M. K Drezner,., F. H. Glorieux, J. A. Kanis, H. Malluche, P. J. Meunier, S. M. Ott, R. R.

Recker (1987). Bone histomorphometry: standarization of nomenclature, symbols, and units,” J Bone

and Mineral Research; Vol. 2(6), p. 595 -610.

105. A. M. Parfitt, C. H. Mathews, A. R. Villanueva, M. Kleerekoper, B. Frame, D. S. Rao (1983).

Relationships between surface, volume and thickness of iliac trabecular bone in aging and in

osteoporosis. Implications for the microanatomic and cellular mechanisms of bone loss. J. Clin. Invest.;

Vol. 72, p.1396-1409.

106. E. Perilli, F. Barufaldi, M. C. Bisi, L. Christofolini, A. Capello (2006). A physical phantom for the

calibration of three-dimensional X-ray microtomography examination. Hournal of Microscopy; Vol. 222,

p. 124-134.

107. F. Peyrin, M. Salome, P. Cloetens, AM Laval-Jeantet, E. Ritman, P. Ruegsegger (1998). Micro-

CT examinations of trabecular bone samples at different resolutions: 14, 7 and 2 micron level. Technol.

Health Care; Vol. 6, p. 391-401.

108. W. Pistoia, B. Van Rietbergen, A. Laib, et al. (2001). High-resolution three-dimensional-pQCT

images can be an adequate basis for-in-vivo mcroFE analysis of bone. J. Biomech. Eng.; Vol. 123, p.

176-183.

109. G. Poumarat, P. Squire (1993). Comparison of mechanical properties of human, bovine bone and

a new processed bone xenograft. Biomaterials; Vol. 14 p. 337.

110. C. S. Rajapakse, J. F. Magland, M. J. Wald, X. S. Liu, X. H. Zhang, X. E. Cuo, F. W. Wehrli

(2010). Computational biomechanics of the distal tibia from high-resolution MR and micro-CT images.

Bone; Vol. 47, p. 556-563.

111. J.Y. Rho, P. Zioupos, J.D. Currey et al. (2002) Microstructural elasticity and regional

heterogeneity in human femoral bone of various ages examined by nanoindentation. Journal of

Biomechanics; Vol. 35, p.189-198.

112. J-Y. Rho, L. Kuhn-Spearing, P. Zioupos. (1998). Mechanical properties and the hierarchical

structure of bone. Medical Engineering & Physics; Vol. 20, p. 92-102.

113. J. Y. Rho, D. Flaitz, V. Swarnakar, R. S. Acharya (1997). The characterization of broadband

ultrasound attenuation and fractal analysis by biomechanical properties. Bone; Vol. 20 p. 497.

114. J. Y. Rho, M. C. Hobatho, R. B. Ashman (1995). Relations of density and CT numbers to

mechanical properties for human cortical and cancellous bone. Med. Eng. Phys.; Vol. 17, p. 347-355.

115. J. J. Rho. M. C. Holbatho, R. B. Ashman (1995). Relations of mechanical properties to density

and CT numbers in human bone. Med. Eng. Phys.; Vol.17(5), p. 347-355.

116. J. Y. Rho (1991). Mechanical properties of human cortical and cancellous bone. PhD

Dissertation, The Univeristy of Texas Southwestern Medical Center at Dallas, Dallas, TX.

117. J. C. Rice, S. C. Cowin, J. A. Bownan (1988) On the dependence of the elasticity and strength of

cancellous bone on apparent density. Journal of Biomechanics; Vol. 21 p. 155-168.

118. B. Van Rietbergen, R. Muller, D. Ulrich, et al. (1999). Tissue stresses and strain in trabeculae of a

canine proximal femur can be quantified from computer reconstructions. J. Biomech.; Vol. 32, p. 443-

451.

119. B. van Rietbergen, S. Majundar, W. Pistoia, D. C. Newitt, M. Kothari, A. Laib, P. Rüegsegger

(1998). Assessment of cancellous bone mechanical properties from micro-FE models based on micro-

CT, pQCT and MR images. Technology and Health Care; Vol. 6, p. 413-420.

120. M. Rossi, F. Casali, D. Romani, M. J. Carabini (2002). 3D micro-CT analysis of cancellous bone

architecture. Developments in X-ray tomography III (ed. By U. Bonse). Proc. SPIE, 4503, p. 349-358.

121. M. Rossi, F. Casali, D. Romani, M. L. Carabini (2001). 3D micro -CT analysis of cancellous bone

architecture. 46th SPIE AnnualMeeting, An Diego (CA), USA; July 29 –Aug 4.

122. T Rüberg - Technische Universität Braunschweig, Centro , 2003 - portal.tugraz.at

123. C. Rubin, AS. Turner, R. Muller et al. (2002). Quantity and quality of trabecular bone in the femur

are enhanced by a strongly anabolic, noninvasive mechanical intervention. J. Bone Miner. Res.; Vol. 17,

p. 349-357.

124. P. Rüegsegger, B. Koller, R. Müller (1996). A microtomographic system for the nondestructive

evaluation of bone architecture. Calcif. Tissue Int.; Vol. 58, p. 24-29.

125. P. Rüegsegger, W. Kalender (1993). A phantom for standardization and quality control in

peripheral bone measurements by pQCT and DXA. Phys. Med. Biol.; Vol. 38, p. 1963-1970.

126. D. J. Sharp, K. E. Tanner, W. Bonfield (1990). Measurement of the density of trabecular bone. J.

Biomechanics; Vol. 23, No. 8 p. 853-857.

127. M. B. Schaffler, D. B. Burr (1988). Stiffness of compact bone: Effects of porosity and density. J.

Biomech.; Vol. 21, p. 13-16.

128. E. Schileo, E. Dall’ Ara, F. Taddei, A. Malandrino, T. Schotkamp, M. Baleani, M. Viceconti (2008).

An accurate estimation of bone density improves the accuracy of subject-specific finite element models.

Journal of Biomechanics; Vol. 41, p. 2483-2491.

129. J.E. Shea and S.C. Miller (2005) Skeletal Function and Structure: Implications for Tissue-

Targeted Therapeutics. Advanced Drug Delivery Reviews; Vol. 57, p.945-957.

130. D I. Singh (1978). The architecture of cancellous bone. Journal of Anatomy; Vol. 127, p. 305-310.

131. S. Tassani, C. Öhman, F. Baruffaldi, M. Baleani, M. Viceconti (2011). Volume to density relation

in adult human bone tissue. Journal of Biomechanics; Vol. 44, p. 103-108.

132. F. Taddei, A. Pancanti, M Viceconti (2004). An improved method for the automatic maping of

computed tomography numbers onto finite element models. Medical Engineering & Physics; Vol. 26, p.

61-69.

133. P. R. Townsend, R. E. Miegel, R. M. Rose, P. Raux, E. L. Radin (1976). Structure and function of

the human patella; the role of cancellous bone. J. Biomed. Mater. Res. Symp.; Vol. 7, p. 605-611.

134. H. S. Tuan, D. W. Hutmacher (2005). Application of micro CT and computation modeling in bone

tissue engineering. Computer-Aided Design; Vol. 37, p. 1151-1161.

135. D. Ulrich, B. Van Rietbergen , A. Laib, et al. (1999). Load transfer analysis of the distal radius

from in-vivo high-resolution CT-imagng. J. Biomech.; Vol. 32, p. 821-828.

136. Y. N. Yeni, C. U. Brown, T. L. Norman (1998). Influence of bone composition and apparent

density on fracture toughness of human femur and tibia. Bone; 22 p.79.

137. J. H. Waarsing, J. S. Day, H. Weinans (2005). Longitudinal micro-CT scans to evaluate bone

architecture. J. ,Musculoskelet Neuronal Interact; Vol. 5(4), p. 310-312.

138. H. H. Waarsing, J. S. Day, H. Weinans (2004). An improved segmentation method for in vivo μCT

imaging. Journal of Bone and Mineral Research; Vol. 19(10), p. 1640-1650.

139. S. Weiner, W. Traub. (1992) Bone Structure: from angstroms to microns. FASEB; Vol. 6, p. 879-

885.

140. D. C. Wirtz, T. Bandorf, F. Portheine, K. Radermacher, N. Schiffers, A. Prescher, D. Weichert, U.

N. Fritz (2003). Concept and development of an orthotropic FE model of the proximal femur. J.

Biomechanics; Vol. 36, p. 289-293.

141. I. Vincent (1993). Morphological Grayscale Reconstruction in Image Analysis Applications and

Efficient Algorithms. IEEE Transactions on Image Processing; Vol. 2(2), p. 176-201.

142. L. Vincent, P. Soille (1991). Watersheds in Digital Spaces: An Efficient Algorithm Based on

Immersions Simulations. IEEE Transactions on Pattern Analysis and Machine Intelligence; Vol. 13(6), p.

583-598.

143. G. P. Vose, A. L. Kubala (1959). Bone strength in relationship to x-ray determined ash content.

Human Biology; Vol. 31, p. 261-270.

144. W. J. Whtehouse, E. D. Dyson (1974). Scanning electron microscope studies of trabecular bone

in the proximal end of the human femur. Journal of Anatomy; Vol. 118, p. 417-444.

145. Z. Yosibash, N. Trabelsi, C. Milgron (2007). Reliable simulations of human proximal femur by

high-order finite element analysis balidated by experimental observation. Journal of Biomechanics; Vol.

40, p.3688-3699.

146. Z. Yosibash, R. Padan, L. Joskowicz, C. Milgron (2006). A CT-based high-order finite element

analysis of the human proximal femur compared to in-vitro experiments. Preprint submitted to Elsevier

Science, p. 1-34.

147. Y. Zhang, Z. He, S. Fan, K. He, C. Li (2008). Automatic Thresholding of Micro-CT Trabecular

Bone Images. International Conference on Biomechanical Engineering and Informatics.

148. Zhi-Feng Sheng, Ru-Chun DaI,Xian-Ping Wu, Yiu-Lin Ma, Kang Xu, Yu-Hai Zhang, Ye-Bin Jiang,

Er-Yuan Liao (2008). The Influence of Center of Rotation on the Assessment of Trabecular Bone

Densitometric and Structural Properties. Annals of Biomedical Engineering; Vol. 36(12), p. 1934–1941.

149. P. Zioupos, R. B. Cook, J. R. Hutchinson (2008). Some basic relationships between density

values in cancellous and cortical bone. Journal of Biomechanics; Vol. 41 p.1961-1968.

150. P. Zioupos, R. Cook, R. M. Aspden, A. M. Coats (2008). Bone quality issues and matix properties

in OP cancellous bone. In: Hammer, J., Nerlich, M., Dendorfer, S. (Eds.), Medicine meets Engineering.

IOS Press, The Netherlands p. 238-245.

151. P. Zioupos (2005) In-Vivo Fatigue Microcracks in Human Bone: Material Properties of the

Surrounding Bone Matrix. European Journal of Morphology; Vol. 42, No. 1/2, p. 31-41

152. P. Zioupos, R. M. Aspden (2000). Density, material quality and quantity issues in OP cancellous

bone. In: Proceedings of the European Society of Biomechanics, Dublin, Ireland, p. 327.

153. Zsolt-Andrei Peter, Françoise Peyrin ( ). Synchrotron Radiation Micro-CT Imaging of Bone Tissue

Theory and Applications of CT Imaging and Analysis

154. www.engin.umich.edu/class/../bonestructure/bonestructure.htm-Cached

155. Structural parameters of trabecular bone measured by the Skyscan Micro-CT.doc.

A1. APPENDIX OF TABLES

Table 1.3A Cortical bone structural organization

Table 1.3B Trabecular bone structural organization

Table 1.5a: Mechanical properties and densities of cancellous bone tissues

Bone (Specimen) Reference

Human-femoral head (8mm diam. cylinder) (Martens, 1983)75

Human proximal femur (8mm diam. cylinder) (Martens, 1983) 75

Human-distal femur (8mm cube) (Kuhn, 1989)45

Human-distal femur(10.3mm diam, 5mm cylinder) (Carter, 1977)14

Human-distal femur(5mm diam/ 7.5mm cylinder) (Odgaard, 1989)99

Human-proximal tibia (7.5:7.5mm cylinder) (Linde, 1989) 75

Human-vertebral body (Keaveny, 1997)60

Monkey-femoral head (5mm diam/6mm cylinder) (Kasra, 1994)57

Cattle-distal femur 95mm diam/8mm cylinder) (Poumarat, 1993)104

Cattle-proximat tibia (5.5 mm diam/8mm cylinder) (Rho, 1997)113

Cattle-proximal humerus (Keaveny, 1997)60

Cattle-vertebral body (6mm diam./7.5mm cylinder) (Swartz, 1991)130

Table 1.5b: Reference sources of the above table

Notice (for Table 3.1.II):

α: ash fraction

Carter and Hayes2:

Reference 16

Gibson9: Reference 42

Current study12: Reference 45

Currey6: Reference 27

Currey7: Reference 26

Schaffler and Burr23:

Reference 127

Vose and Kubala24:

Reference 140

Table 3.4.I. Average values (Standard deviation) of stereological parameters for

scanning/reconstruction voxel size groups

Table 3.1.II: Power law constants (constants (exponents) for

BV/TV and α

Table 3.4.II: Resolution/voxel dependency of morphological parameters of bone

(cortical, trabecular) (24)

Notice for Table 3.4.II:

Interpretation for morphological parameters:

BV/TV = bone volume fraction; CaV/TV = cortical porosity

BS/BV = bone surface to volume ratio; CaS/CaV = canal surface to volume ratio

TbTh = mean trabecular thickness; CaDm = mean canal diameter; TbSp = mean trabecular

separation

CaSp = mean canal separation; TbN = trabecular number; CaN = canal number; DA = degree

of anisotropy DA = degree of anisotropy

Interpretation for References:

Muller et al. 1996: Reference 92

Kothari et al. 1998: Reference 69

Peyrin et al. 1998: Reference 107

Kim et al. 2004: Reference 67

Table 3.4.III: Material and structural properties of trabecular bone which would be

influenced by center displacement(148)

Dose (µg/kg) Number of Samples

0 7

0.03 6

0.1 8

0.3 7

1 6

Table 4.1; Number of treated samples with dose

Standard Name Mineral (%)

Antler 44

Dugong– Trabecular 49.5

Dugong –Laminar 67

Dugong – Filler 73

Rostrum 90

Table 4.3: Mineral content of five standards

Tasks(Duties) and application states of CT-Agent

Tasks (Duties) Receiving CT data acquired by X- Inspect when performing

CT Acquisition and writing these files to disk

Managing a queue of CT reconstruction jobs

Synchronising the above activities and CT-Pro reconstruction

setup to best use of processing and storage resources

Application states Reconstructing (There are jobs within the reconstruction job

queue and they are free to be reconstructed.)

Receiving files

(X-inspect acquires a CT-data set or CT-Pro requires to

reconstruct a slice of the volume.)

Resource busy (Any reconstruction job will be paused and

the Receiving files sate can not be entered.)

(The reconstruction job will be re-started when CT-Pro has

finished reconstructing slices and the Resource Busy state is

exited.)

Idle

(The default state for CT-Agent. There is no job to be

reconstructed.)

Paused

(Jobs within the reconstruction queue will not be

reconstructed.)

Table 5.1: CT-Agent (Duties and Application states)

Option Icon Position

Toggle fullscreen mode Toggle fullscreen state of respective window

Focus selected object Focus on selected object

Reset zoom Resets zoom of two-dimensional (2-D window)

Zoom in Zoom into 2-D window

Zoom out Zoom out of 2-D window

Lock slice position Locks the slice position

Set brightness Set brightness

Toggle transformation

mode

Toggle the mode in which you can move/rotate

the selected object

Original display mode Select original color mode to display the slice

image in its original, brightness and contrast

settings

Color mode /

color and opacity mode

Swich between “Color mode” and “Color and

opacity mode”

Create/replace clipping

plane

Create/replace a clipping plane based on the

current slice view

Navigation plane Display/hide the navigation planes

Configure

Define the rotation axis of the rotation window

Show scale bar Show/hide the scale bar

4/2/1 pixels per ray Specify the amount of downsampling in the 3D

window

Rotate Enables rotating the object in 3D

Move Enables moving the object in 3D

Show tripod Show/hide the tripod in the 3-D view

Show box

Show/hide the box in 3-D view

Front, back, left, right, top,

bottom

Select one of the default camera view s for the 3-

D window

Freeze rendering Enable/disable the rendering process

Center and

focus camera

Centers the selected objects and makes them

fully visible

Table 5.2.A: Workspace window controls

Option Icon Description

New Create a new project

Open Open an existing project

Save Save the current project under the same file

name

Save as… Save the current project and specify a filename

Merge project.. Merges objects from a different project file into

the current scene

Save object.. Save the currently selected object into a

dedicated project file

Pack and go.. Gathers all project data dependencies and

saves them into a compact project file

Import (+ submenu) Import an object (several format options)

Export (+submenu) Export an object (several format options)

Save image(s).. Save selected views (i.e. slice views) as

images (several format options)

Save AVI/image stack Save selected visible items of the scene into an

image stack or .avi file

Print image(s) Print selected views (e.g. slice views) as

images

Quit

Exit the application

Table 5.2.B: File menu options

Option Icon Description

Rotate Enables rotating the object in 3-D

Move Enables moving the object in 3-D

Clipping plane Creates a clipping plane for the selected object

Aligned clipping box Creates an aligned clipping box for the selected

object

Clipping polyline 3D Creates a clipping polygon for the selected object

Create point/directional

spot light

Creates an additional light source of the

respective type for the selected object

Determine surface

Defines the material boundary

Remove surface

determination

Deletes the surface determination information

Registers the selected object against another

object in the scene (i.e. superposition of the two

objects)

Unregister object Deletes the registration information

Merge polygon objects

Merges several polygon objects into a new

polygon object

Group Render Objects Combines same or combatible type objects which

then behave like a single object

Ungroup Render Objects Dissolves a group back into separate objects

Draw Draws the selected grey value directly to the

voxels in freehand mode

Pick color Samples the selected voxel for its grey value and

shares the value with the “Grey value selection”

tool. Only available in the 2-D Windows

Fill Fills the selected ROI with the current grey value

Properties

Displays object properties

Table 5.2.C: Object menu options

Option Icon Description

Selection modes

(+ submenu)

Choose a mode by

which to create a new

selection. Includes

simple drawing tools as

well as advanced modes

with automatic adaption

Selection mode>Draw Paint into selection

mask using a circular or

spherical pencil

Selection mode>Rectangle Create a rectangular

selection in the chosen

2-D view

Selection mode>Round

Rectangle

Create a rectangular

selection with rounded

corners in the chosen 2-

D view

Selection mode> Ellipse Creates an elliptical

selection in the chosen

2-D view

Selection mode> Polyline Creates a polygonal

selection in the chosen

2-D view (discrete

control points)

Selection mode>Lasso Creates a polygonal

selection in the chosen

2-D view (continuous

control points)

Selection mode>Region

growing

Creates a selection

using a region growing

(flooding) algorithm

Table 5.2.D: Select menu options

Option Icon Description

Selection mode>Adaptive

rectangle

Find the edges of an object inside a

rectangular research region using the

edges of the rectangular region as

initial guess

Selection mode>Adaptive

polyline

Fits an arbitrarily shaped polygon to

the edges of within an object

Selection mode>Adaptive line Fits a line of a given width to the

edges of within an object

Selection mode>Polyline 3-D Creates a polygonal section in the 3-D

view

Selection mode> Crack

segmentation

Grows inside dark crack-structures

based on an existing region of interest

Selection mode>Erode/Dilate Perform a morphological operation on

the ROI: Expand or contract it.

ROI from surface Create a region of interest from the

surface determination

ROI from defect mask Create a region of interest based on

the defect mask

ROI from wall thickness mask Create a region of interest based on

the wall thickness mask

Enable ROI render settings Add/Remove render settings to/from

the current region of interest

Add ROI to ROI Adds a region of interest to another

region of interest

Subtract ROI from ROI Subtracts a region of interest from

another region of interest

Intersect ROI with ROI Performs a logical AND between the

selected regions

Split ROI Split the selected region of interest

Merge ROIs Combine region of interest to one

region of interest

Invert ROI Invert the selected region of interest

Extract ROI Extract the selected region of interest

as a new object

Import ROI template(s) Import regions of interest

Export ROI template(s)

Export regions of interest

Table 5.2.D(supplement): Select menu options (continue (or supplement))

Option Icon Description

New defect detection Analyze the select objects for defects

New nominal/actual

comparison

Calculate a nominal/actual comparison

New wall thickness

Analyze the wall thickness of the selected

object

New analysis annotations

Create annotations for the selected

analysis

Table 5.2.E: Analysis menu options

A2. APPENDIX OF FIGURES

A.2.I FIGURES DERIVED FROM FROM LITERATURE STUDIES

Figure 1.2; Tertiary diagram showing the relationships and variation between the

different constituent components of bone and the variation occuring within nature

(Ref. 7)

Figure 1.3A: Macrostructure of cortical bone derived from the femoral head of a 69yr

Osteoarthritic male (Ref. 23)

Figure 1.3B: Macrostructure of cancellous bone derived from the femur of a 92yr old

male (Ref. 88)

Figure 1.4: The internal structure of cortical bone in a dog's left femur

Figure 2.1 Dapp vs Dmat in 4 different cohorts (175 samples, 66 OP (osteoporotic), 12

OA (osteoarthritic)) of donors(152, 150) with least squares and 95% prediction intervals for

the data

Comment-(Apart from a small number of OA data (arrows), the general trend does not aim for

the customarily assumed hypothetical end-point of the process (encircled) at a nominal bone

matrix density of 2.2 g cm-3.)(140)1

Figure 2.2.1: Apparent vs Material density for all samples (triangles) produced from the

same femur in both cortical and cancellous regions. (samples having Dapp>1.3 g cm-3

are encircled)(140)

Figure 2.2.2: Material density vs mineral content (ash weight over wet mineralized

weight) for all samples (solid triangles) and those with Dapp>1.3 g cm-3 (encircled)

(Least squares linear regression with its 95% confidence interval and 95% prediction

interval for the data; Dmat=0.70±0.028Min%, R2=0.77)(140)

Figure 2.2.3.a: Plot of E vs Dapp for all samples (solid triangles) and those with

Dapp>1.3 g cm-3 (encircled) (140)

Figure 2.2.3.b: Plot of E vs Dmat for all samples (solid triangles) and those with Dapp>1.3

g cm-3 (encircled) (140)

Figure 2.2.3.c: Plot of E vs BV/TV Dmat for all samples (solid triangles) and those with

Dapp>1.3 g cm-3 (encircled) (140)

Figure 2.2.4: Mineral content vs BV/TV for all samples

(solid triangles) and those with Dapp>1.3 g cm-3 (encircled) (140)

Figure 3.1.I.B: Plotting bone surface area density as a function of bone volume

fraction(20)

Figure 3.1.IIa: Distribution of bone volume and ash fraction(45)

Figure 3.1.IIb: Regression analyses of ρapp/ρash for a)pooled trabecular and cortical

tissue

b) trabecular and cortical groups separated, c) for the trabecular groups obtained from

bovine femurs and human femurs separated, d) for the cortical specimens obtained

from bovine femurs and the trabecular specimens from bovine femurs(128)

Figure 3.1.IIA: Regression analysis for ash density versus bone volume fraction for a)

and b) cases a.) pooled data b) splitted plot for trabecular and cortical tissue(131)

Figure 3.1.IIB: Regression analysis for ρash versus BV/TV (for inter-site, intra-site

analysis)(131)

Figure 3.1.IIC: Variation of TMD due to changes in BV/TV for cortical and trabecular

specimens(131)

Figure 3.2.I.A: S-I modulus-density correlations from present and previous studies

(denstities of other studies have been converted to ash density (65)

Notice (for Figure 3.2.I):

Anderson et al.1: Reference 2

Ashman et al.2: Reference 2

Carter and Hayes3: Reference 16

Hvid et al.5: Reference 51

Linde and Hvid6: Reference 77

Linde et al.7: Reference 76

Linde et al.8: Reference 78

Odgaard et al.10: Reference 101

Present Study: Reference 65

Figure 3.2.I.B: S-I strength-density correlations from present and previous studies

(densities of other studies have been converted to ash density) (117)

Notice (for Figure 3.2.II):

Anderson et al.1: Reference 2

Ashman et al.2: Reference 2

Carter and Hayes3: Reference 16

Hvid et al.5: Reference 51

Linde and Hvid6: Reference Error! Reference source not found.

Linde et al.7: Reference 76

Linde et al.8: Reference 78

Odgaard et al.10: Reference 101

Present Study: Reference 65

Figure 3.2.II: The elastic modulus is plotted against the apparent density of the distal

femur

Notice for Figure 3.2.II.A.

1: Anterior-posterior direction

2. Medial-lateral direction

3. Superior-inferior direction

Figure 3.3.II. A: Generation of a p-FE model(145)

Figure 3.3.II.B: Hounsefield Unit (HU) numbers(134)

Figure 3.4.I: Representative 2D micro-CT cross-sections at matched magnifications

(scaled

without interpolation)(24)

Lower left insert represent original image sizes relative to 5µm scan image.

Scanned datasets along top row: 5 (left), 10 (center), and 15 (right) µm voxel sizes.

Artificially degraded datasets along bottom row: 10 (left), 20 (center), and 40 (right) µm

voxel sizes.

Figure 3.4.II: Comparison of Elastic modulus (along principal directions) computed

from micro-CT images at 25µm isotropic voxel size with the corresponding values

obtained after resampling the imkages at lower resolutions(110)

Figure 4.1: Samples in the mould with resin

Figure 3A: The effect of global and local thresholds on in vitro and in vivo scans of

whole bones(138)

A. Absolute difference in the estimation of cortical bone and trabecular bone with

respect to the locally segmented data set as a function of the global threshold. At the optimum

(grey arrow), there is a difference between the volumes of the locally and globally segmented

data set

B. Example of in vitro scan, showing a grey-value cross-section (B1) that is

C. Example of in vitro scan, showing a grey-value cross-section (B1) that is

segmented with the local method (B2) and the global method using the optimal threshold (B3)

D. Example of in vivo scan, showing a grey-value cross-section (C1), both

segmented with the local method (C2).

(It can be pointed out that the overestimation of the cortical bone and the subchondral bone in

the globally segmented cross-sections.)

Figure 3.3.I: Reconstructed images of a hand cross-section obtained with different

preprocessing approaches(9) It seems that the trabecular structure of bones derived

from fingers creates a texture.

Figure 3.3.II: a. Representative CT-image of a hand

Result obtained from grey-scale mathematicl morphology image a) reconstruction(9)

Figure 3.3.III: region-based segmentation using the watershed(Bossart A., 5) concept

Figure 3.3.III(a):Original shape

Figure 3.3.III (b): Markers after thresholding of the image

Figure 3.3.III (c): Gradient of the image in Figure 3.3.III (a)

Figure 3.3.III (d): Implementation of region growing algorithm on Figure 3.3.III (a)

A.2.II FIGURES DERIVED FROM OUR MICRO-CT SCANNING ANALYSIS

IMAGES OBTAINED BY USING CT-PRO SOFTWARE

Figure 5.1.A; Open file dialog in CT Pro

Figure 5.1.B: First projection of the data set displayed on Image tab

Figure 5.1.C. Processing Tabs on User Interface

Figure 5.1.D: Center of rotation controls

Figure 5.1.E: Slice displayed throughout calculation of the center of rotation

Figure 5.1.F: Set up controlsΕικόνα 2Figure 5.1.F:Set up controls

Figure 5.1.G: Image displayed on Set up tab

Figure 5.1.H: Volume controls

Figure 5.1.I: Volume projections viewed by 0o and 90o

Figure 5.1.K: Image displayed on Volume tab

Figure 5.1.L: Calibration controls

IMAGES OBTAINED BY USING VGSTUDIOMAX SOFTWARE

Figure 5.2.B: Reconstructed volume displayed on VGLStudiomax panel

Figure 5.2A: Reconstructed slice displayed on VGLStucdiomax panel

Figure 5.2C: Scene Tree & Volume Rendering

Figure 5.2.D: Navigation cursor on 2-D window

Figure 5.2.E: Automatic surface determination results

Figure 5.2F: Define material by example area

A3. APPENDIX OF EQUATIONS

Bone

eral

BMC min

= (Equation 1.5.2)

where: eral

mmin is bone mineral mass

Bone

m is weight of the dry bone

real

app

−= 1(Equation 1.5.3)

where:

app

is apparent density

real

is real density

Data

December 2019

Calibration panel.bmp

Data

December 2019

Automatic surface detemination.bmp

Data

December 2019

Centre of Rotation panel

Data

September 2017

Surface determination

Data

September 2017

aCTIVATE NAVIGATION CURSOR

Data

September 2017

Opening vgi file

Data

September 2017

Volume panel (Picture 1)

Data

September 2017

Calibration panel

Data

September 2017

Set up-Beam Hardening reduction

Data

September 2017

Noise reduction

Data

September 2017