ENHANCING SEGMENTATION APPROACHES FROM GC-OAAM AND MTANN TO FUZZY K-C-MEANS

Article (PDF Available)inInvestigación clínica 59(1):129-138 · March 2018with 25 Reads
Cite this publication
Abstract
Medical Image Segmentation is an activity with huge handiness. Biomedical and anatomical data are made simple to acquire because of progress accomplished in computerizing picture division. More research and work on it has improved more viability to the extent the subject is concerned. A few techniques are utilized for therapeutic picture division, for example, Clustering strategies, Thresholding technique, Classifier, Region Growing, Deformable Model, Markov Random Model and so forth. This work has for the most part centered consideration around Clustering techniques, particularly k-implies what's more, fluffy c-implies grouping calculations. These calculations were joined together to concoct another technique called fluffy k-c-implies bunching calculation, which has a superior outcome as far as time usage. The calculations have been actualized and tried with Magnetic Resonance Image (MRI) pictures of Human cerebrum. The proposed strategy has expanded effectiveness and lessened emphasis when contrasted with different techniques. The nature of picture is assessed by figuring the proficiency as far as number of rounds and the time which the picture takes to make one emphasis. Results have been dissected and recorded. Some different strategies were surveyed and favorable circumstances and hindrances have been expressed as special to each. Terms which need to do with picture division have been characterized nearby with other grouping strategies.
Invest Clin 59(1): 129-138 , 2018
Abstract:
Medical Image Segmentation is an activity with huge handiness. Biomedical and anatomical data are
made simple to acquire because of progress accomplished in computerizing picture division. More
research and work on it has improved more viability to the extent the subject is concerned. A few tech-
niques are utilized for therapeutic picture division, for example, Clustering strategies, Thresholding
technique, Classifier, Region Growing, Deformable Model, Markov Random Model and so forth. This
work has for the most part centered consideration around Clustering techniques, particularly k-implies
what's more, fluffy c-implies grouping calculations. These calculations were joined together to
concoct another technique called fluffy k-c-implies bunching calculation, which has a superior outco-
me as far as time usage. The calculations have been actualized and tried with Magnetic Resonance
Image (MRI) pictures of Human cerebrum. The proposed strategy has expanded effectiveness and
lessened emphasis when contrasted with different techniques. The nature of picture is assessed by figu-
ring the proficiency as far as number of rounds and the time which the picture takes to make one
emphasis. Results have been dissected and recorded. Some different strategies were surveyed and
favorable circumstances and hindrances have been expressed as special to each. Terms which need to
do with picture division have been characterized nearby with other grouping strategies.
Keywords: Graph Cut Method, Active Contours Model, Geodesic Graph Cut Method, Graph-Cut
Oriented Active Appearance Model (GC-OAAM), Massive Training Artificial Neural Network
(MTANN), Fuzzy-K-C-Means Segmentation Method.
ENHANCING SEGMENTATION APPROACHES FROM
GC-OAAM AND MTANN TO FUZZY K-C-MEANS
Christo Ananth1, S.Aaron James2, Anand Nayyar3, S.Benjamin Arul4, M.Jenish Dev5
1College of Engineering, AMA International University, Kingdom of Bahrain
2Electrical & Electronics Section, Ibri College of Technology, Sultanate of Oman
3Graduate School, Duy Tan University, Da Nang, Vietnam
4Department of EIE, Jeppiaar Engineering College, Chennai, India
5Department of ECE,DMI Engineering College, Aralvaimozhi, India
130
Invest Clin 59(1): 129-138 , 2018
Resumen
Segmentación de imagen médica es una actividad con gran manejabilidad. Los datos biomédicos
y anatómicos son fáciles de adquirir debido al progreso logrado en la computación de la división
de imágenes. Más investigación y trabajo en él ha mejorado más la viabilidad en la medida en
que se trata el tema. Se utilizan algunas técnicas para la división de imágenes terapéuticas, por
ejemplo, estrategias de agrupamiento, técnica de umbral, clasificador, región en crecimiento,
modelo deformable, modelo aleatorio de Markov, etc. Este trabajo, en su mayor parte, se centra
en las técnicas de agrupación en clústeres, especialmente k-implica lo que es más, mullido
c-implica agrupar cálculos. Estos cálculos se unieron para crear otra técnica llamada mullida
k-c-implica el cálculo de agrupamiento, que tiene un resultado superior en cuanto al uso del
tiempo. Los cálculos se actualizaron y se probaron con imágenes de imagen de resonancia mag-
nética (IRM) del cerebro humano. La estrategia propuesta ha ampliado la efectividad y ha dismi-
nuido el énfasis cuando se contrasta con diferentes técnicas. La naturaleza de la imagen se evalúa
calculando la competencia en cuanto al número de rondas y el tiempo que toma la imagen para
hacer un énfasis. Los resultados han sido diseccionados y registrados. Se examinaron algunas
estrategias diferentes y se han expresado circunstancias y obstáculos favorables como especiales
para cada una. Los términos que tienen que ver con la división de imágenes se han caracterizado
cerca con otras estrategias de agrupación.
Palabras clave: Método de corte de gráfico, Modelo de contornos activos, Método de corte de
gráfico geodésico, Modelo de aspecto activo orientado a corte de gráfico (GC-OAAM), Red
neuronal artificial de entrenamiento masivo (MTANN), Método de segmentación de Fuzzy-K-
C-Means.
MEJORANDO LOS ENFOQUES DE SEGMENTACIÓN
DE GC-OAAM Y MTANN A FUZZY K-C-MEANS
I INTRODUCTION
I. INTRODUCTION
Restorative imaging is a vital apparatus for
finding and treatment arranging today. Ano-
ther proposed technique for completely
programmed handling structures is given in
light of Geodesic Graph-cut Active Contour
calculations. A predicate is characterized for
estimating the confirmation for a limit
between two districts utilizing Geodesic
Graph-based portrayal of the picture. The
calculation is connected to picture division
utilizing two various types of nearby neigh-
borhoods in building the chart. The real
issue with Graph-Cut approach is the
mistaken determination of Liver Region
with shading like client's scrawls being
recognized as a tumor area. Results can be
enhanced by utilizing the proposed new
procedure in light of Geodesic Graph-Cut
strategy. This framework has focused on
finding a quick and intelligent division tech-
nique for liver and tumor division. In the
preprocessing stage, the CT picture process
ENHANCING SEGMENTATION APPROACHES FROM GC-OAAM AND MTANN TO FUZZY
K-C-MEANS
131
Investigacion Clinical 59(1). 2018
Christo Ananth et al.
is persisted with mean move channel and
factual thresholding technique for lessening
handling region with enhancing identification
rate. Second stage is Liver Segmentation; the
liver area has been divided utilizing the calcu-
lation of the proposed strategy. In the
following stage, Tumor division likewise took
after similar advances. At long last the liver
and tumor locales are independently divided
from the PC tomography picture. In this
undertaking, a programmed life structures
division technique is proposed. This techni-
que viably joins the Active Appearance
Model, Live Wire and Graph Cut plans to
misuse their integral qualities. It comprises of
three primary parts: show building, introduc-
tion, and outline. For the introduction
(acknowledgment) part, utilize a pseudo
methodology and section the organs cut by cut
by means of the OAAM technique. The
reason for introduction is to give unpleasant
question restriction and shape requirements
for a last GC strategy, which will deliver
refined depiction.
It is smarter to have a quick and powerful
strategy than a moderate and more precise
method for introduction. One of the qualities
of our strategy is to portion the organ amid the
procedure of physically outlining the limits.
For single-protest division, worldwide opti-
mality is ensured. Multi-shape GC OAAM
division strategies have actualized on guts
MRI and edge recognition on liver picture. It
enhances the question depiction time than the
current technique. This approach for division
of restorative pictures can help in the best
possible identification of the district of
premium and furthermore can be exceptiona-
lly useful for specialist's findings, therapeutic
educating, learning and research. The
remarkable element of the calculation is that it
can consequently pick an ideal shape earlier
from among different priors at each voxel by
limiting the proposed sub-secluded vitality
work. The minimization is per-shaped in a
consecutive way by the combination move
calculation that uses the QPBO min-cut calcu-
lation. Multi-shape GCs are turned out to be
more valuable than single-shape GCs. Subse-
quently, the division techniques are approved
by computing factual measures. The false
positive (FP) is lessened and affectability and
specificity enhanced by numerous MTANN.
The similar execution investigation is done
between GC-OAAM and MTANN yields. The
execution of the MTANN is straightforwardly
connected to the false positives. The multi
determination deterioration/structure strate-
gies with two down/up-testing steps enabled
MTANNs to help a 28.8-by-28.8 mm square
locale. Major focus towards time reduction
needed for the image prefiltering by emplo-
ying more powerful computers.Satisfactory
results obtained in tumor delineations may be
exploited for future improvement regarding
the detection of cysts. The motivation of future
enhancement of the project is finding alternati-
ve method of segmentation for improving
results and reducing execution of time for a
test image. In this method have testing with
different set of images like as other part of the
body CT and MR images and if a possible to
getting 3D image data’s for any one of the
object can test by the knowledge of this
method. First plan is to optimize the number of
shape priors used in the multi-shape graph
cuts.
II. EXISTING SYSTEM
132
ENHANCING SEGMENTATION APPROACHES FROM GC-OAAM AND MTANN TO FUZZY
K-C-MEANS
Invest Clin 59(1): 129-138 , 2018
The Existing System utilized Automatic and
Semiautomatic strategies for Liver division
which gives manual depictions of liver shapes
and tumors on indicative pictures. In any case, it
expends additional time and it produces inhe-
rent low reproducibility. 2 Main Approaches
were essentially included. They are Intensity-
based and Model-based techniques. Power
construct strategy is situated in light of Thres-
holding and Morphological sifting. Demonstra-
te construct approach is situated in light of
dynamic forms display. In any case, here there
is no legitimate setting of handling steps and
methodology and it requires manual instate-
ment in view of learning database or on UIs.
The Existing framework utilized Interactive
picture division by means of versatile weighted
separations which don't unequivocally consider
and precisely limit protest limits. Additionally
the Curvature consistency strategy gives arch
limiting method to smooth limits. Be that as it
may, it doesn't utilize an edge segment to
restrict edges and it devours additional time. In
the past work, Graph-cut approach was utilized
which is expressly utilized as a part of edge-
finding and utilized as locale displaying
segments. In this proposed technique, same
division strategy for liver was connected and
for its inner neurotic structures.
Abdel-Massieh.N.H. et al. [1] investigated the
completely programmed and proficient method
for liver division from stomach CT pictures
which depends on Fast Marching to section
liver districts from Multi-cut Spiral Computed
Tomographic Images. The paper proposed a
completely programmed strategy to fragment
liver districts from MSCT. In any case, connec-
tions between's neighbor cut pictures were
gotten to get beginning fronts. At that point an
adjusted quick walking was connected to
engender the fronts until stop rule was fulfi-
lled. The regions included by the proliferated
fronts were the liver areas. Finish liver locales
of one case could be gotten from each cut utili-
zing comparable approach one by one. This
paper assessed nearby and worldwide data
which gave exact liver limit and there was
brilliant Correlation between neighbor cut
pictures. Anyway it doesn't function admirably
when liver tissues have disparate powers with
adjoining organs volumetric estimation and 3D
perception of liver.
Ben-Dan.I. et al. [2] assessed the Liver Tumor
division technique in CT pictures utilizing
probabilistic strategies which depend on
Chan-Vese strategy (Energy based Segmenta-
tion) utilizing power probability proportion
test. Initial an underlying histogram and mea-
surable appropriation capacities are made, and
from them another picture is made where, in
each voxel, a weighted capacity is connected
as per the likelihood of the voxel dim level.
Next, the dynamic form technique on the new
picture is utilized, where the dynamic shape
development depends on the minimization of
changes between the liver tumor and its nearest
neighborhood. Here mix of strategies for
earlier investigation and vitality based division
is utilized. Vitality construct division is situa-
ted in light of Active forms technique without
edges and Active shape and division utilizing
geometric Probability Density Energy work.
Effective numerical strategies were produced
for veins division and for liver division. The
Segmentation strategy utilized has created
better and less clamor touchy outcomes.
Anyway vessels division isn't legitimately
approved and liver parceling isn't completed
appropriately which are the significant weak-
nesses.
133
Investigacion Clinical 59(1). 2018
Christo Ananth et al.
Boykov.Y. et al. [3] explored a fundamental
answer for surface development PDEs by
means of geo-cuts technique which models
slope streams of forms and surfaces. While
standard variational strategies (e.g. level sets)
register nearby interface movement in a diffe-
rential form by assessing neighborhood shape
speed by means of vitality subordinates, surface
advancement PDEs was illuminated by
expressly evaluating vital movement of the
entire surface. An improvement issue was detai-
led specifically in light of a necessary portrayal
of inclination stream as a microscopic move of
the (entire) surface giving the biggest vitality
diminish among all moves of equivalent size.
This issue can be proficiently unraveled utili-
zing late advances in calculations for worldwi-
de hyper surface improvement. Specifically, the
geo-cuts strategy was utilized that utilizations
thoughts from indispensable geometry to speak
to consistent surfaces as cuts on discrete
diagrams. The subsequent interface develop-
ment calculation is approved on nearly 2D and
3D illustrations like run of the mill shows of
level-set strategies. This technique can register
inclination streams of hyper surfaces as for a
genuinely broad class of consistent useful and it
is adaptable regarding separation measurements
on the space of shapes/surfaces. The calculation
creates an opportune arrangement of slices
comparing to inclination stream of a given
shape. This technique is definitely not another
execution of level set strategies yet rather an
option numerical strategy for developing inter-
faces. Our technique does not utilize any level
set capacity to speak to forms/surfaces. Rather,
it utilizes a certain form/surface portrayal
through geo-cuts. As the level set technique,
this approach handles topological changes of
the developing interface. A basic approach was
you can fertilize the sperm and the egg outside
the uterus, and inject the egg into a fertility-
producing uterus. Because under normal
conditions, the host womb accepts the egg cell
for its own work, a legal review of this issue
can be considered in discussing the nature of
the contract for the use of the uterus, and under
the title "Rent of the womb".
It is clear that in this method, the woman after
giving birth gave the child to her parents and
takes her wages against it. But since there is a
parenting relationship between the child and
the suckling mother as well as the owner of the
egg, which mother is the mother of the child,
which is one of the two, is a discussion that
will be discussed in its place [13].
2. Traditional succession of the uterus:
Another type proposed in the succession of
pregnancy is the traditional succession of the
womb and, in other words, "substitution with
artificial fertility." In this way, based on an
agreement with the succeeding mother, the
male sperm of the infertile couple is artificially
transferred to his uterus, which after delivery
gives birth to a child and surrenders after
giving birth. In fact, in this case, the mother's
egg is transfused by the male sperm, which is
the father of the decree, and the mother
succeeds to carry her fertilized ovum with the
sperm of the decree; therefore, the mother of
the substitute (minor substitute mother) and
The father will also have a genetic link with the
child; Because the suckling mother's egg
carries fertility with the fertilization of the
male sperm by intrauterine fertilization or the
extrauterine fertilization with the male sperm,
and the suckling mother carries an embryo
derived from her egg fertilized by the sperm of
the father, and the infertile wife, the mother An
applicant will be considered and will have no
134
ENHANCING SEGMENTATION APPROACHES FROM GC-OAAM AND MTANN TO FUZZY
K-C-MEANS
Invest Clin 59(1): 129-138 , 2018
fashions are bodily inspired techniques. Deli-
neation of an object boundary in an photo is
executed by setting a closed curve or floor close
to the desired boundary then an iterative relaxa-
tion technique is allowed to be undergone. inner
forces are computed from within the curve or
floor to preserve it easy for the duration of the
deformation. Outside forces are typically deri-
ved from the photograph to pressure the curve
or surface closer to the preferred feature of
interest.
III. PROPOSED SYSTEM
Cluster Evaluation or Clustering is the task of a
firm of perceptions into subsets (alluded to as
groups) so perceptions inside the equivalent
group are tantamount in a couple of involve-
ment. Bunching is a strategy for unsupervised
contemplating, and an ordinary system for mea-
surable certainties investigation utilized as a
part of numerous fields, for example, gadget
picking up learning of, measurements mining,
test notoriety, picture examination, insights
recovery, and bioinformatics. Bunching calcu-
lations and the classifier technique are perhaps
in trademark yet grouping does now not utilize
tutoring insights rather they repeat between
fragmenting the photo and describing the habi-
tations of each class. therefore they are in some
other case named unsupervised techniques. In a
vibe, grouping systems teach themselves the
use of the accessible realities . three for the most
part utilized grouping calculations are the
alright means, the shaggy C-way calculation,
and the desire augmentation (EM) calculation.
The alright way bunching calculation groups
realities through iteratively processing a normal
power for each eminence and dividing the
picture by methods for ordering every pixel
inside the radiance with the closest propose.
A. Fuzzy C-means Clustering
Due to the upsides of Magnetic Resonance
Imaging (MRI) over other demonstrative
imaging, the overall population of investigates
in clinical photograph division relate to its
utilization for MR pictures, and there are an
assortment of systems to be had for MR photo
division. among them, fluffy division systems
are of monster favors, because of the reality
they could keep an awesome arrangement
additional records from the first photograph
than intense division procedures. particularly,
the Fuzzy C-Means (FCM) calculation, alloca-
te pixels to fluffy bunches without marks. not
at all like the hard bunching strategies in some
other case called alright means grouping which
weight pixels to have a place totally with one
brilliance, FCM lets in pixels to have a place
with two or three groups with shifting levels of
enrollment. because of the extra adaptability,
the fluffy C-approach bunching set of princi-
ples (FCM) is a delicate division technique that
has been utilized essentially for division of MR
pix applications of late. in any case, its essen-
tial dangers include its computational unpre-
dictability and the truth that the general execu-
tion corrupts remarkably with expanded
commotion.
Fuzzy c-means (FCM) is a method for
bunching which lets in a single snippet of data
to have a place with at least 2 groups. In
various word, each factor has a level of having
a place with groups, as in fluffy good judg-
ment, instead of having a place totally with 1
bunch. as a result, focuses on the very edge of
a bunch can be in the group to a lesser recogni-
tion than focuses inside the focal point of
bunch. Fluffy c-strategy has been an absolutely
essential gadget for picture handling in
bunching objects in a picture. in the 70's,
mathematicians conveyed the spatial term into
136
ENHANCING SEGMENTATION APPROACHES FROM GC-OAAM AND MTANN TO FUZZY
K-C-MEANS
Invest Clin 59(1): 129-138 , 2018
Where is a chosen distance measure between a
data point and the cluster centre , is an indicator
of the distance of the n data points from their
respective cluster centres. K-means is a simple
algorithm that has been adapted to many
problem domains. It is a good candidate for
extension to work with fuzzy feature vectors.
IV. SIMULATION RESULTS
A. Operation Mode
K-means requests that the client determines the
quantity of groups before the division initiates.
Subsequently, the quantity of bunches is foreor-
dained. The k-means strategy considered here is
working in view of hues contained by the pictu-
re. The quantity of groups determined by the
client must relate to the quantity of shading. It
isn't important to have the pre-information of
the quantity of hues contained by the picture in
light of the fact that there is arrangement made
for re-contributing the quantity of groups. Grea-
test number of conceivable hues accommodated
is 9 since most pictures may have as much as
5-6 hues. It is conceivable to have a picture
whose hues are more than this range, conse-
quently the arrangement for more hues. When
k-implies gets to the finish of the bunches indi-
cated it stops. Fuzzy C-Means changes over a
hued picture into dim scale before beginning
the division. That is it sections utilizing dim
scale. In the event that the picture inputted is a
non-hued it will in any case portion it dissimilar
to the k-means which just fragments a hued
picture. Generally, Fuzzy C-means repeats in
light of the quantity of bunches it runs over on
the picture being considered. Not at all like
K-means, the fuzzy c-means will restore the
quantity of bunches after the division has been
finished. Consequently the number groups is
roughly the quantity of cycles. Fuzzy K-C-
Means is a strategy produced from both Fuzzy
c-means and k-means however it conveys a
greater amount of Fuzzy c-means properties
than that of k-means. Fuzzy k-c-means takes a
shot at dim scale pictures like Fuzzy c-means,
produces an indistinguishable number of
emphasess from in Fuzzy c-means. In view of
the tried pictures k-means seems, by all
accounts, to be speedier than Fuzzy c-means
while sometimes Fuzzy c-means likewise
gives off an impression of being quicker than
k-means. Though both Fuzzy c-means and
k-means are contending as far as time, Fuzzy
k-c-means has been modified to produce a
similar number of emphasis with Fuzzy
c-means with a quicker activity time. That is
Fuzzy k-c-means is quicker than both Fuzzy
c-means and k-means. The contention in time
between Fuzzy c-means and k-means is accep-
ted to account from the properties of the pictu-
re under thought, the proficiency of the machi-
ne on which the techniques are tried.
B. Accuracy
As far as exactness, the number cycle is put
into thought. The more the emphasess the more
the precision. The emphasis that k-means can
perform depend to a great extent on the quanti-
ty of hues contained by a picture which make
its iterative capacity constrained not at all like
that of Fuzzy c-means and Fuzzy k-c-means
which portion in light of the quantity of cycles
or bunches contained in a picture. Resulting to
this, k-means is less exact than the other two
techniques.
Investigacion Clinical 59(1). 2018
Invest Clin 59(1): 129-138 , 2018
references:
[1] Alizadeh, Mahdi, (2006), Legal Status of
the Use of Alternative Uterus, Journal of
Specialist Theology and Law, p. 19, p. 179.
Tehran, Iran.
[2] Frank J. Langu. Fertilization (fertility),
translation of Sepehri Hoori, Tehran University
Pr005), Introduction to IVF and the necessity
of using
137 Christo Ananth et al.
138
ENHANCING SEGMENTATION APPROACHES FROM GC-OAAM AND MTANN TO FUZZY
K-C-MEANS
The strategy with the most astounding cycle
esteem and sections inside the briefest timefra-
me takes the more exactness. For this situation
Fuzzy k-c-means ought to have been conside-
red however with clear perception GC-OAAM
is slower than Fuzzy k-c-means along these
lines Fuzzy k-c-means takes the most astoun-
ding precision.
C. CONCLUSION
Medical Image Segmentation is an activity
with huge handiness. Biomedical and anatomi-
cal data are made simple to acquire because of
progress accomplished in computerizing pictu-
re division. More research and work on it has
improved more viability to the extent the
subject is concerned. A few techniques are
utilized for therapeutic picture division, for
example, Clustering strategies, Thresholding
technique, Classifier, Region Growing, Defor-
mable Model, Markov Random Model and so
forth. This work has for the most part centered
consideration around Clustering techniques,
particularly k-implies what's more, fluffy
c-implies grouping calculations. These calcu-
lations were joined together to concoct another
technique called fluffy k-c-implies bunching
calculation, which has a superior outcome as
far as time usage. The calculations have been
actualized and tried with Magnetic Resonance
Image (MRI) pictures of Human cerebrum.
The proposed strategy has expanded effective-
ness and lessened emphasis when contrasted
with different techniques. The nature of picture
is assessed by figuring the proficiency as far as
number of rounds and the time which the pictu-
re takes to make one emphasis. Results have
been dissected and recorded. Some different
strategies were surveyed and favorable
circumstances and hindrances have been
expressed as special to each. Terms which need
to do with picture division have been characte-
rized nearby with other grouping strategies.
D. SCOPE FOR FUTURE ENHANCE-
MENT
In a future work, we will research Hidden
Markov Random Fields with Expected Maximi-
zation for bigger scale organize estimate, and
consider the effective answer for the expanded
effectiveness and lessened emphasis in a system
with high portability. It is additionally a promi-
sing future work to match the accuracy and
object delineation time of the existing system
for multi-shape structures.
REFERENCES
[1] Abdel-massieh.N.H., Hadhoud.M.M., and
Moustafa.K.A.,(2010) “A fully automatic and
efficient technique for liver segmentation from
abdominal CT images,” in Proc. 7th Int. Conf.
Inform. Syst. (INFOS),pp. 1–8.
[2] Ben-Dan.I. and Shenhav.E.,(2008) “Liver
Tumor segmentation in CT images using pro-
babilistic methods,” in Proc. 11th Int. Conf.
Med. Image Comput. Comput. Assisted Inter-
vention, MICCAI’08,, New York, pp. 1–11.
[3] Boykov.Y. and Kolmogorov.V.,(2003)
“Computing geodesics and minimal surfaces
via graph-cuts,” in Proc. 4th Int. Conf. Comput.
Vision, Nice,pp. 26–33
[4] Lim.S.J., Jeong.Y.Y., and Ho.Y.S.,(2006)
“Automatic liver segmentation for volume mea-
surement in CT images,” J Vis. Commun.
Image Represent., vol. 17, pp. 860–875.
[5]Yong Yang, Shuying Huang,(2007) “Image
Segmentation By Fuzzy C-Means Clustering
Algorithm With A Novel Penalty Term” Com-
puting and Informatics, Vol. 26, 2007, 17–31.
Invest Clin 59(1): 129-138 , 2018
This research hasn't been cited in any other publications.
  • Conference Paper
    As a mean for liver investigation, abdominal CT images have been widely studied in the recent years. Processing CT image includes the automatic diagnosis of liver pathologies, such as detecting lesions and following vessels ramification, and its 3D volume rendering. The first step in all these studies is the automatic liver segmentation. This paper presents a fully automatic method to segment the liver from abdominal CT data with no interaction from user. A statistical model-based approach is used to distinguish roughly liver tissue from other abdominal organs. It is followed by applying force-driven optimized active contour (snake) in order to obtain a smoother and finer liver contour. The new segmentation technique has been evaluated on fifteen datasets, by comparing the automatically detected liver contour to the liver boundaries manually traced by an expert. Tests are reported on 15 datasets and promising result shows that sensitivity and specificity for automatic liver segmentation are 95% and 99% respectively.
  • Article
    Computed tomography (CT) images have been widely used for diagnosis of liver disease and volume measurement for liver surgery or transplantation. Automatic liver segmentation and volume measurement based on the segmentation are the most essential parts in computer-aided diagnosis for liver CT as well as computer-aided surgery. However, liver segmentation, in general, has been performed by outlining the medical image manually or segmenting CT images semi-automatically because surface features of the liver and partial-volume effects make automatic discrimination from other adjacent organs or tissues very difficult. Accordingly, in this paper, we propose a new approach to automatic segmentation of the liver for volume measurement in sequential CT images. Our method analyzes the intensity distribution of several abdominal CT samples and exploits a priori knowledge, such as CT numbers and location of the liver to identify coherent regions that correspond to the liver. The proposed scheme utilizes recursively morphological filter with region-labeling and clustering to detect the search range and to generate the initial liver contour. In this search range, we deform liver contour using the labeling-based search algorithm following pattern features of the liver contour. Lastly, volume measurement is automatically performed on the segmented liver regions. The experimental measurement of area and volume is compared with those using manual tracing method as a gold standard by the radiological doctors, and demonstrates that this algorithm is effective for automatic segmentation and volume measurement method of the liver.
  • Article
    Full-text available
    Liver tumors segmentation is an important prerequisite for planning of surgical interventions. For clinical applicability, the segmentation approach must be able to cope with the high variation in shape and gray-value appearance of the liver. We present a fully automatic 3D segmentation method for the liver tumors from contrast-enhanced CT data. The method consists of two main stages. First an initial histogram and statistical distribution functions are created, and from them a new image is created where, in each voxel, a weighted function is attached in accordance with the probability of the voxel grey level. Next, we use the active contour method on the new image, where the active contour evolution is based upon the minimization of variances between the liver tumor and its closest neighborhood.
  • Article
    Geodesic active contours and graph cuts are two standard image segmentation techniques. We introduce a new segmentation method combining some of their benefits. Our main intuition is that any cut on a graph embedded in some continuous space can be interpreted as a contour (in 2D) or a surface (in 3D). We show how to build a grid graph and set its edge weights so that the cost of cuts is arbitrarily close to the length (area) of the corresponding contours (surfaces) for any anisotropic Riemannian metric.