MRI protocols for whole-organ assessment of the knee in osteoarthritis

Article (PDF Available)inOsteoarthritis and Cartilage 14 Suppl A(supplement 1):A95-111 · February 2006with32 Reads
DOI: 10.1016/j.joca.2006.02.029 · Source: PubMed
Abstract
One of the critical challenges in developing structure-modifying therapies for arthritis, especially osteoarthritis (OA), is measuring changes in progression of joint destruction. Magnetic resonance imaging (MRI) offers considerable promise in this regard. Not only can MRI quantify articular cartilage volume and morphology with high precision and accuracy, but it can also examine several other important articular components, and thus offer a unique opportunity to evaluate the knee and other joints as whole organs. On December 5 and 6, 2002, OMERACT (Outcome Measures in Rheumatology Clinical Trials) and OARSI (Osteoarthritis Research Society International), with support from various pharmaceutical companies listed at the beginning of this supplement, held a Workshop for Consensus on Osteoarthritis Imaging in Bethesda, MD. The aim of the Workshop was to provide a state-of-the-art review of imaging outcome measures for OA of the knee to help guide scientists and pharmaceutical companies who want to use MRI in multi-site studies of OA. Applications of MRI were initially reviewed by a multidisciplinary, international panel of expert scientists and physicians from academia, the pharmaceutical industry and regulatory agencies. The findings of the panel were then presented to a wider group of participants for open discussion. The following report summarizes the results of these discussions with respect to MRI acquisition techniques for whole-organ assessment of the knee in OA. The discussion reviews the selection and qualification of imaging sites for clinical trials, designing imaging protocols for whole-organ assessment of OA, and key considerations in image quality (IQ) control and data management.
3 Figures
MRI protocols for whole-organ assessment of the knee in osteoarthritis
C. G. Peterfy M.D., Ph.D.y*, G. Gold M.D., Ph.D.z, F. Eckstein M.D.xk,
F. Cicuttini M.D.{, B. Dardzinski Ph.D.# and R. Stevens M.D.yy
y Synarc Inc., San Francisco, CA, USA
z Department of Diagnostic Radiology and Magnetic Resonance Systems Research Laboratory,
Stanford University, Stanford, CA, USA
x Institute of Anatomy, Paracelsus Private Medical University, Salzurg, Austria
k Chondrometrics GmbH, Munich, Germany
{ Department of Epidemiology and Preventive Medicine, Monash University Medical School,
Alfred Hospital, Prahran, Australia
# Imaging Research Center, Children’s Hospital Medical Center, Cincinnati, OH, USA
yy Hoffman-LaRoche Inc, Nutley, NJ, USA
Summary
One of the critical challenges in developing structure-modifying therapies for arthritis, especially osteoarthritis (OA), is measuring changes in
progression of joint destruction. Magnetic resonance imaging (MRI) offers considerable promise in this regard. Not only can MRI quantify ar-
ticular cartilage volume and morphology with high precision and accuracy, but it can also examine several other important articular compo-
nents, and thus offer a unique opportunity to evaluate the knee and other joints as whole organs. On December 5 and 6, 2002, OMERACT
(Outcome Measures in Rheumatology Clinical Trials) and OARSI (Osteoarthritis Research Society International), with support from various
pharmaceutical companies listed at the beginning of this supplement, held a Workshop for Consensus on Osteoarthritis Imaging in Bethesda,
MD. The aim of the Workshop was to provide a state-of-the-art review of imaging outcome measures for OA of the knee to help guide sci-
entists and pharmaceutical companies who want to use MRI in multi-site studies of OA. Applications of MRI were initially reviewed by a mul-
tidisciplinary, international panel of expert scientists and physicians from academia, the pharmaceutical industry and regulatory agencies. The
findings of the panel were then presented to a wider group of participants for open discussion. The following report summarizes the results of
these discussions with respect to MRI acquisition techniques for whole-organ assessment of the knee in OA. The discussion reviews the se-
lection and qualification of imaging sites for clinical trials, designing imaging protocols for whole-organ assessment of OA, and key consider-
ations in image quality (IQ) control and data management.
ª 2006 OsteoArthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
Key words: Osteoarthritis, Knee, MRI, Whole organ, Protocol.
Image analysis, regardless of the sophistication and talent
of the readers or the power of the image-processing and
analysis software used, can only be as good as the quality
of the original images acquired. Use of improper imaging
technique or the presence of serious artifacts can render
image data useless. Good image analysis therefore begins
with good image acquisition and careful quality control. Per-
forming this properly on a multi-site or multi-national scale
can be extremely challenging, and requires special exper-
tise and systems not found in mainstream clinical practice.
Multi-site clinical trials require imaging techniques that
are widely available, reproducible at different sites and sta-
ble over time. They must also cause minimum patient dis-
comfort and non-compliance. This is distinct from the
common focus of university research on cutting-edge tech-
nology, which may have only single-site applicability.
Proper consideration of the factors relevant to multi-site
research facilitate study start-up, accelerate patient
recruitment, decrease patient drop-out and missing data,
and minimize sources of variability that undermine mea-
surement precision and statistical power.
On December 5 and 6, 2002, OMERACT (Outcome Mea-
sures in Rheumatology Clinical Trials) and OARSI (Osteo-
arthritis Research Society International), with support from
various pharmaceutical companies listed at the beginning
of this supplement, held a Workshop for Consensus on
Osteoarthritis Imaging in Bethesda, MD. The overall aim
of the Workshop was to provide a state-of-the-art review
of imaging outcome measurement in osteoarthritis (OA) to
help guide scientists and pharmaceutical companies who
want to use Magnetic resonance imaging (MRI) in multi-
site studies of knee OA. Applications of MRI were initially
reviewed by a multidisciplinary, international panel of expert
scientists and physicians from academia, the pharmaceuti-
cal industry and regulatory agencies. The panel was co-
chaired by Charles Peterfy, M.D., Ph.D. (Synarc, Inc., San
Francisco, CA, USA) and Roy Altman, M.D. (University of
Miami, Miami, FL, USA
a
) and also included Deborah Bur-
stein, Ph.D. (Harvard-MIT, Cambridge, MA, USA), Flavia
Cicuttini (Epidemiology, Monash University, Prahran,
*Address correspondence and reprint requests to: Charles
G. Peterfy, Synarc, Inc., 575 Market Street, 17th Floor, San
Francisco, CA 94105, USA. Tel: 1-415-817-8901; Fax: 1-415-
817-8999; E-mail: charles.peterfy@synarc.com
Received 10 May 2004; revision accepted 26 February 2006.
a
Dr Altman is currently at UCLA, CA, USA.
OsteoArthritis and Cartilage (2006) 14, A95eA111
ª 2006 OsteoArthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.joca.2006.02.029
International
Cartilage
Repair
Society
A95
Australia), Gary Cline, Ph.D. (Biometrics and Statistical Sci-
ences, Proctor & Gamble Pharmaceuticals, Madison, OH,
USA), Philip Conaghan, M.B.B.S., F.R.A.C.P. (Rheumatol-
ogy, Leeds University, Leeds, UK), Bernard Dardzinski,
Ph.D. (MRI Physics, University of Cincinnati, Cincinnati,
USA), Felix Eckstein, M.D. (MRI analysis, Ludwig-Maximili-
ans-Universita
¨
t, Mu
¨
nchen, Germany
b
), David Felson, M.D.,
M.P.H. (Rheumatology, Boston University, Boston, MA,
USA), Garry Gold, M.D., Ph.D. (Radiology, Stanford Univer-
sity, Stanford, CA, USA), Benjamin Hsu, Ph.D. (GlaxoS-
mithKline, Research Triangle Park, NC, USA
c
), Marissa
Lassere, M.B.B.S., Ph.D., F.R.A.C.P. (Epidemiology, St.
George Hospital, Kogarah, Australia), Stefan Lohmander,
M.D., Ph.D. (Orthopaedics, University of Lund, Lund,
Sweden), Jean-Pierre Raynauld, M.D. (Rheumatology,
University of Montreal and Arthrovision, Montreal, PQ, Can-
ada), Randall Stevens, M.D. (Hoffman-LaRoche Inc, Nutley,
NJ, USA), Saara Totterman, M.D., Ph.D. (Virtual Scopics,
Pittsford, NY, USA), James Witter, M.D. (Food and Drug
Administration (FDA), Washington, DC, USA), and Thasia
Woodworth, M.D. (Pfizer, Groton, CT, USA
d
). The panel
met in New Orleans, LA on October 29, 2002 prior to the
Workshop in Bethesda to define a preliminary set of MRI
features to include in whole-organ assessment of the
knee
1
and to review the relative strengths and weaknesses
of various imaging protocols for multi-feature, multi-site
MRI. The findings of the panel were presented to the partic-
ipants of the Workshop in Bethesda for open discussion.
This report summarizes the results of these discussions
with respect to MRI acquisition for whole-organ assessment
of the knee in OA.
Selecting and qualifying imaging sites
Selecting imaging sites for a clinical trial is a complex
matter. Important considerations include:
1. The type and quality of imaging equipment available at
the site and its compatibility with the equipment at the
other sites.
2. The competence, motivation, reliability and clinical tri-
als experience of the imaging technologists at the sites.
3. Proximity to desirable clinical investigators.
4. Patient safety, comfort and convenience.
5. Availability and ease of scheduling, as trial imaging
competes with clinical imaging slots.
6. The process for transferring images between the site
and the central radiology service.
7. The cost of imaging.
These factors must be considered in the light of the spe-
cific scientific, regulatory and strategic objectives of the trial,
the proposed method of image analysis, and the compe-
tence, experience and systems compatibility of any central
radiology service that will supervise the imaging, manage
the image data and perform the image analyses.
The degree to which imaging equipment across the mul-
tiple sites included in a clinical trial must be standardized
depends on the type of measurements that will be made,
and the ability of the central radiology service to deal with
multi-vendor image data. Ideally, all equipment, software
platforms and upgrade schedules at all of the sites
throughout the duration of the study should be identical,
but this is rarely feasible. Knowing what deviations from
this ideal can be tolerated without compromise to the scien-
tific, regulatory and strategic integrity of the study requires
considerable sophistication and experience. Sometimes,
for example, equipment changes between serial visits are
unavoidable. This can result in variations in signal-to-noise
ratio (S/N), spatial homogeneity, and a variety of other pa-
rameters that can affect the study results. In the worst case
such changes are systematic, creating an off-set in the
data between visits. In such cases, specialized phantoms
designed to detect and measure deviations in specific
parameters may be able to correct the technical variations.
Phantoms can also be used to cross-calibrate different
imaging devices at the study sites. This enables the use
of multi-vendor imaging equipment and thereby increases
the pool of imaging sites applicable to a study. Phantoms
are also used to monitor the stability of imaging equipment
over time. Deviations in performance detected with these
image-acquisition aids can be corrected by feedback to
the imaging site or occasionally by post-processing the im-
ages with specialized corrective software using quantitative
information provided by the phantom.
How the data will be transferred to the image reading and
management service is also an important consideration.
The cost of mailing hardcopy images can be high and
increases with the number of sites included. In contrast,
unit cost of electronically transferring electronic images
decreases with the number of sites networked. Only a few
sites around the world are currently networked properly to
allow electronic image transfers, but the number is increas-
ing rapidly as teleradiology becomes more widespread.
Alternatively, electronic image data can be transferred on
a variety of inexpensive and relatively high storage capacity
media, such as magnetic tape or optical disc. Despite the
fact that most imaging devices are currently formatted
according to the Digital Imaging and Communications in
Medicine (DICOM) 3.0 standard
2
, managing and analyzing
multi-vendor image data still requires specialized software.
For example, files may be stored on digital linear tapes in
DICOM 3.0 format, but the media itself may be proprietary.
Motivated, competent imaging technologists with direct
experience conducting clinical trials are essential to the suc-
cess of any study. Since it is unlikely that imaging for clinical
trials will ever be more than only a marginal interest for
mainstream radiology practice, clinical imaging sites usually
find it difficult to support protocols that deviate substantially
from their clinical routines. Accordingly, in almost all cases
imaging technologists must be given special training in
order to perform the imaging protocol accurately and con-
sistently. This can be accomplished either through individ-
ual-site training, or through centralized training sessions at
investigator meetings, and supplemented with detailed
study manuals, videos and interactive web based instruc-
tional programs. Test runs of the imaging protocol and
use of any image-acquisition aids that simplify technolo-
gists’ work can also be extremely helpful. These image-
acquisition aids include not only phantoms for calibrating
and potentially correcting image quality (IQ), as mentioned
above, but also positioning devices that ensure reproduc-
ible imaging on serial examinations.
Often the proximity of the imaging site to a desirable
clinical investigator is a critical factor. Convenience for the
patient is also important. Scheduling imaging time for clini-
cal trials can be difficult, as research cases compete with
clinical cases, and it is usually the latter that are the sites’
main priority. Limited availability of imaging time slots can
b
Dr Eckstein is currently at Paracelsus Private Medical University
Salzburg, Austria and Chondrometrics GmbH, Munich, Germany.
c
Dr Hsu is currently at Centocor, Philadelphia, PA.
d
Dr Woodworth is no longer at Pfizer.
A96
C. G. Peterfy et al.: MRI protocols for whole-organ assessment
slow a study considerably, and repeated cancellation and
rescheduling can lead to patient drop-out. In some studies,
imaging hubs fed by several clinical recruitment sites are
used to reduce the total number of imaging sites needed.
This can reduce cost and variability, and elevate the level
of protocol complexity that can be supported, but these
potential benefits must be balanced carefully against patient
inconvenience and scheduling/capacity problems.
The cost of imaging can vary considerably from site to site
and from country to country. International currency
exchange rates may need to be considered in global trials.
Additionally, some sites offer unit pricing, whereas others
charge according to imaging time used. Since subject prep-
aration and examination set-up activities can constitute as
much as a third of the total MRI examination time, how
different imaging sites account for these times can have
a significant impact on pricing. Set-up activities include final
screening for contraindications to MRI, discussions about
the procedure, changing the subject into the imaging gown
and removing any metallic items, etc. that can interfere
with the examination, having the subject void before the ex-
amination, placing the subject into the scanner, and position-
ing the knee in the imaging coil, as well as any coil
repositionings that may be needed during the examination,
e.g., when switching from one knee to the other in bilateral
examinations. Minimizing set-up time and patient anxiety
depend on the experience and skill of the technologist, but
at a minimum, 10e15 min of patient set-up time are typically
required in addition to the actual time required to run the
pulse sequences in order to complete an MRI examination
of the knee. Repeat examinations because of protocol viola-
tions or poor IQ, regardless of whether they were the result
of inadequate training, fundamental lack of competence of
the site, or non-compliance of the patient, can add significant
cost to a study. Accordingly, imaging cost must not be con-
sidered in isolation of the competence, experience, reliability
and convenience of the imaging site in question.
Designing imaging protocols
When designing any imaging protocol, one must decide
two critical factors: (1) which articular structures and
features will be included in the assessment, and (2) what
measurement method(s) must the images support for
each structure or feature. These choices dictate how best
to acquire the images. Additionally, one must balance the
needs for IQ, convenience and cost against the scientific,
regulatory, logistical and budgetary constraints of the study.
Key IQ parameters include full anatomical coverage, appro-
priate image orientation, sufficient contrast-to-noise ratio,
spatial resolution and, spatial and signal homogeneity,
and absence of technical artifacts. These factors are inter-
dependent and must be optimized in an integrated fashion
rather than in isolation. Other key constraints are patient
tolerance and safety, and the availability of requisite imag-
ing hardware and software.
The consensus of the Workshop participants was that the
most important articular features to include in whole-organ
assessment of knee OA with MRI were articular cartilage,
osteophytes, bone marrow abnormality, meniscal integrity,
synovial effusion/tissue, anterior and posterior cruciate liga-
ments, and medial and lateral collateral ligaments. Of these,
Fig. 1. Critic al anatomical coverage for arti cular cartilage. Lateral
sagittal section of the knee showing the minimum anatomical cov-
erage required to include the entire patellar, femoral and tibial car-
tilage surfaces.
2D SE
3D GRE
365 µ
365 µ
3000 µ
365 µ
365 µ
700 µ
CB
A
Fig. 2. Voxel dimensions determine plane dependency. Pixel signal intensity on an MRI is the average signal of all structures included within
the corresponding voxel. The effect of this partial-volume averaging on delineation of cartilage interfaces is illustrated in panel B, which is
a magnification of the region of femoral cartilage indicated in A. Partial-volume averaging is greatest along the longest dimension of a voxel,
which is typically the through-plane direction. This is an important consideration in selecting the imaging plane. S/N ultimately limits voxel size
and, therefor e, spatial resolution, so 2D SE images require larger voxels than 3D GRE images do. Thi s is achi eved by lengthening the
through-plane dimension (slice thickness) of 2D voxels (C). Accordingly, plane selection has a greater impact on 2D images.
A97
Osteoarthritis and Cartilage Vol. 14, Supplement A
articular cartilage morphology (e.g., volume, thickness, car-
tilaginous/denuded surface area, etc.) was considered to be
the most important. Accordingly, adequate anatomical cov-
erage for most studies should include the entire patellar,
femoral and tibial cartilages (Fig. 1). This may be accom-
plished with several different scans, but all of these cartilage
surfaces should be covered by the protocol. How well the
cartilage within the field of view (FOV) of any scan is delin-
eated depends on the imaging coil, magnet field strength,
pulse sequence parameters, plane of section and voxel di-
mensions selected. Voxel size is determined by multiplying
slice thickness by the area of the in-plane subdivisions of
the image (pixel size), which in turn is determined by divid-
ing the FOV by the image matrix. The smaller the voxel, the
greater the spatial resolution. However, as voxel size
decreases, so does the number of hydrogen protons within
each voxel and therefore the S/N of the image. High-resolu-
tion imaging requires sufficient S/N to support the small
voxel size. S/N can be increased in a number of ways,
including shortening echo time (TE) (less T2 decay), in-
creasing repetition time (TR) (more T1 recovery), increasing
the number of excitations (NEX) averaged, imaging at
higher field strength (greater longitudinal magnetization),
or utilizing specialized coils which reduce noise (small sur-
face coil, quadrature coil, phased array of many small
coils)
3,4
. Specialized sequences, such as three-dimensional
(3D) gradient-echo (GRE), particularly using steady-state
free precession (SSFP) technique, also provide greater
S/N, which can be leveraged to achieve higher spatial re-
solution or to shorten overall imaging time
5
.
There are several factors to be considered when select-
ing the plane of section and voxel size. Most of these are
based on the geometry of the cartilage surface of interest.
For a given voxel size, articular cartilage is optimally delin-
eated when the shortest dimension of the voxel is aligned
with the shortest dimension of the cartilage, i.e., when ori-
ented perpendicular to the cartilage surface (Figs. 2e5).
Unless a voxel is isotropic (all dimensions equal), the
longest voxel dimension is typically in the through-plane
direction (slice thickness). Curvatures of cartilage in the
through-plane direction are, therefore, more subject to
partial-volume averaging and poor delineation. Accordingly,
different regions of cartilage are variably depicted in differ-
ent planes (Figs. 3e5).
Patellar cartilage is optimally imaged in the axial plane.
Sagittal images section most of the patellar cartilage
obliquely, and therefore offer reduced IQ, but the delinea-
tion may be adequate for some purposes. Indeed, sagittal
images depict the superior and inferior poles of the patella
better than axial images do, so a combination of axial and
sagittal images is ideal (Table I). Coronal images, on the
other hand, section the articular surface of the patella
enface and therefore are not suitable for assessing patellar
cartilage.
Tibial cartilage is well imaged in the coronal plane, as
most of the curvature of this cartilage is along the tibial
spines, which is perpendicular to this plane. Sagittal voxels
are well oriented to delineate most of the tibial surfaces, but
section the cartilage over the tibial spines obliquely. As for
patellar cartilage, this may not be a significant limitation
for some purposes. Axial images, however, do not depict
the tibial cartilage adequately.
Of all of the articular surfaces, the femoral cartilage has
the most complex geometry and accordingly poses the
greatest challenge to single-plane imaging. Most of the cur-
vature of the femoral cartilage is in the sagittal plane, and
therefore is well delineated with sagittal voxels; although,
some of the anterior and posterior portions of the central
surfaces (the region beneath the femoral notch and that
articulates with the tibia when the knee is in extension)
are sectioned obliquely. Coronal sections provide excellent
images of this central femoral cartilage, but section the
anterior and posterior surfaces of the femur obliquely or
enface, and therefore inadequately, unless very thin sec-
tions are used. Axial images provide good delineation of
the superior aspects of the anterior and posterior femoral
cartilages, but section most of the remainder of the femoral
cartilage, particularly the important central surfaces, enface
or obliquely.
All things considered, therefore, sagittal images probably
offer the best single-plane global assessment of articular
cartilage in the knee. Ideally, however, multiple planes
Fig. 3. Sagittal plane dependency of arti cular cartilage. (A) Lateral section of a sagittal image set provides good delineation of both the
patellofemoral and femorotibial cartilage interfaces. Axial (B) and coronal (C) localizer images show orientation of through-plane voxel dimen-
sion (between vertical lines) relative to articular cartilage surfaces during sagittal imaging. Panel B shows that except at the patellar ridge and
trochlea r groove the patellar and anterior femoral cartilages are oriented obliquely within the sagittal voxel s. Partial-volume averaging is,
accordingly, greater along these obliquely sectioned articular surfaces. Panel C shows that cartilage over the medial tibial spine (arrow)
and the lateral margin of the central medial femur in this example are also obliquely sectioned on sagittal images.
A98
C. G. Peterfy et al.: MRI protocols for whole-organ assessment
should be examined, as coronal and to a lesser extent axial
images provide important complementary information and
even superior delineation of certain surfaces.
Another important consideration, particularly for longitudi-
nal investigations, is reproducible orientation and positioning
of the voxels on serially acquired images. These variations
between examinations can be corrected to some extent after
the acquisitions, using a variety of image registration
techniques. All of these post-processing techniques, how-
ever, involve some loss of spatial resolution. This cost
in IQ can be minimized by maximizing the reproducibility
of voxel orientation at the time of follow-up imaging. This
is accomplished by standardizing the positioning of
the knee in the imaging coil and carefully aligning the
imaging planes with pre-specified anatomical landmarks
(Figs. 6e8).
In addition to spatial resolution, delineation of articular
cartilage depends on its contrast with adjacent structures.
Since many different structures (synovial fluid, synovial
tissue, intra-articular fat, menisci, opposing cartilage sur-
faces, the posterior capsule of the knee) abut the cartilage,
discriminating all of these interfaces with a single pulse
sequence is challenging (Fig. 9). Because of this and the
planar considerations discussed above, multiple pulse
sequences are usually included in the MRI protocol. It is
important, therefore, to consider a protocol as a whole, rec-
ognizing both the synergies and potential redundancies
among its constituent pulse sequences.
One important interface is that between the articular car-
tilage and the subchondral bone plate. Since MRI depicts
bone tissue as a signal void,
e
pulse sequences, such as
T2-weighted fast spin-echo (FSE),
f
which depict cartilage
with low signal intensity, offer reduced contrast at this in-
terface. Since the subchondral plate is a relatively thin
structure and adjacent marrow fat provides high contrast
with bone and cartilage, this may not be a serious
limitation for some indications. However, if T2-weighted
FSE is used in conjunction with fat suppression (FS) or se-
lective water excitation (WE), and the contrast between
marrow and the subchondral bone plate is consequentially
diminished, maintaining positive signal throughout the
cartilage can be more important. This can be accom-
plished by using intermediate-weighted (IW) rather than
T2-weighted FSE. FS similarly lowers contrast between
cartilage and subchondral osteophytes, cartilage and the
joint capsule posteriorly, and cartilage and intra-articular
adipose tissue, such as Hoffa’s fat pad anteriorly. Again,
intermediate-weighting of FSE images can reduce this
problem. FS is useful in cartilage imaging because it elim-
inates chemical-shift phenomenon, which is a spatial mis-
registration of signal from protons in fat relative to those
from water. This artifact distorts morphology at fatewater
interfaces in the frequency-encoding direction, and can
simulate cartilage thinning or thickening in certain loca-
tions. Chemical-shift effect can be minimized by decreas-
ing readout bandwidth, but is eliminated all together by
spectral FS or WE. FS also augments T1 contrast with
3D GRE techniques, such as spoiled gradient recalled
(SPGR) or fast low-angled shot (FLASH), which provide
the highest cartilageebone contrast. Fat-suppressed 3D
SSFP techniques
6e9
also provide high cartilageebone
contrast, and may offer additional advantages in terms of
S/N and imaging speed.
Another important interface is that between the cartilage
surface and synovial fluid. Cartilageefluid contrast is high
with all of the techniques described above, but generally
greater with T2-weighted FSE and fat-suppressed 3D
SSFP than with fat-suppressed T1-weighted 3D GRE.
However, because of its higher spatial resolution, fat-sup-
pressed T1-weighted 3D GRE typically outperforms two-
dimensional (2D) FSE with respect to delineating cartilage
morphology. T2-weighted 2D FSE, on the other hand, is su-
perior to fat-suppressed T1-weighted 3D GRE for detecting
collagen-matrix abnormalities, which show decreased T2
relaxation and therefore elevated signal on T2-weighted or
IW images. Dual-echo steady-state (DESS) and SSFP
Fig. 4. Coronal plane dependency of articular cartilage. (A) Central section of a coronal image set provides good delineation of the central
femorotibial cartilage interfaces. Axial (B) and sagittal (C) localizer images show orientation of through-plane voxel dimension (between ver-
tical lines) relative to articular cartilage surfaces during coronal imaging. Panels B and C show that the patellofemoral and posterior femoral
cartilages are oriented enface within co ronal voxels, resulting in maximum partial-volume averaging and poor delineation of these articular
surfaces unless very thin sections are used. In contrast, coronal voxels are optimally oriented for delineating tibial cartilage. Except for the
central-most portion of femoral cartilage articulating with the tibia (brackets in C), much of the central femoral cartilage is obliquely oriented
in coronal voxels.
e
MRI depic ts cortical and trabecul ar bone as a sign al void be-
cause of the relative lack of hydrogen protons in this tissue.
f
FSE is also referred to as turbo SE.
A99
Osteoarthritis and Cartilage Vol. 14, Supplement A
techniques may also provide information about cartilage
matrix integrity (Fig. 9), although the performance of these
techniques relative to that of FSE has not been systemati-
cally studied. Conventional T1-weighted spin-echo (SE)
offers very poor cartilageefluid contrast, but excellent delin-
eation of osteophytes and meniscal abnormalities (Fig. 10)
(see below).
Cartilageemeniscus contrast is usually high with most
pulse sequences that offer good contrast at other cartilage
interfaces. However, delineating the interface between
abutting cartilage surfaces (cartilageecartilage contrast) is
often difficult with fat-suppressed T1-weighted 3D GRE
(Fig. 11). Because the articular surfaces in the knee are
relatively incongruent, this often represents only a small
portion of the total cartilage surface in this joint. However,
poor cartilageecartilage contrast can be a major limitation
in congruent joints such as the hip. Some investigators
have overcome this problem by imaging the hip during
mild joint distraction
10
.
Marginal osteophytes are relatively well delineated with
most pulse sequences that are also useful for imaging
cartilage. Axial images are useful for delineating medial
and lateral osteophytes along the anterior and posterior
margins of the femur, and are essential for imaging medial
and lateral osteophytes of the patella (Fig. 5). Sagittal
images are necessary for delineating superior and inferior
osteophytes of the patella, superior osteophytes of the an-
terior and posterior femur and anterior and posterior osteo-
phytes of the tibia (Fig. 12). Central marginal osteophytes,
which most closely correspond to the osteophytes depicted
on frontal radiographs of the knee, are best delineated on
coronal MRI.
Bone marrow abnormalities associated with OA
11e13
are
most sensitively demonstrated with fat-suppressed T2-
weighted FSE images and short-tau inversion recovery
(STIR) images. One limitation of fat-suppressed T2-
weighted images is that local magnetic field heterogeneities
near irregularly shaped anatomy, such as the patella, or
near metal can result in areas of failed FS, which can mimic
marrow abnormality (Fig. 13). Distinguishing between true
marrow abnormality and failed FS, particularly in the setting
of prepatellar bursitis can be challenging, but in contrast to
these biological processes, FS failure does not respect nor-
mal tissue boundaries. Other common technical artifacts in
knee MRI, including metallic artifacts, vascular pulsation,
and aliasing, or ‘‘wrap’’ artifact, are discussed below and
illustrated in Figs. 14e16.
Marrow abnormality can also be seen on heavily T1-
weighted images but not as sensitively as with fat-
suppressed T2-weighted images. GRE techniques, even
with robust FS or WE, are notoriously insensitive to marrow
abnormality because of trabecular magnetic susceptibility,
or T2*, effects (Fig. 17). Bone marrow abnormality can be
semiquantitatively scored or quantified volumetrically. How-
ever, the feathery, ill-defined margins of this feature make it
difficult to segment accurately, particularly in the presence
of heterogeneous FS. Validating the technical accuracy of
putative methods for quantifying the volume of marrow
abnormality is equally challenging, because of the difficulty
associated with correlating the measurements with
appropriate histopathological standards.
Subarticular bone cysts and bone attrition are also char-
acteristic, albeit late, features of OA that are well delineated
with conventional MRI. Fat-suppressed or water-excitation
3D GRE techniques used for imaging articular cartilage
are also excellent for delineating subarticular cysts and
bone attrition. The insensitivity of these techniques for
bone marrow abnormality is actually advantageous for de-
lineating bone cysts, as the two features often occur adja-
cent to each other and can be difficult to distinguish in
some cases ( Fig. 17).
Meniscal tears are most sensitively delineated with short-
TE MRI pulse sequences, such as T1-weighted SE or IW
FSE
14
. The tissue contrast provided by fat-suppressed
T1-weighted 3D GRE sequences optimized for cartilage
imaging, and by long-TE pulse sequences, such as T2-
weighted FSE, is relatively insensitive for meniscal tears.
Novel pulse sequences, such as DESS
6
and other SSFP
techniques show promise for assessing meniscal integrity
6
,
but experience in this respect is still limited. The majority of
meniscal tears are demonstrable on sagittal images, but
a small percentage of body-segment tears are visible only
Fig. 5. Axial plane dependency of articular cartilage. (A) Midpatellar section of an axial image set provides good delineation of the patellar
cartilage and the superior aspects of the anterior and posterior femoral cartilages. Coronal (B) and sagittal (C) localizer images show orien-
tation of through-plane voxel dimension (between vertical lines) relative to articular cartilage surfaces during axial imaging. Panels B and C
show that the tibial and central femoral cartilages are oriented enface within axial voxels, resulting in maximum partial-volume averaging and
poor delineation of these articular surfaces unless very thin sections are used. Note the well-delineated marginal osteophytes of the medial
and lateral margins of the patella and the anterior and posterior femur on this axial fat-suppressed T1-weighted 3D GRE image.
A100
C. G. Peterfy et al.: MRI protocols for whole-organ assessment
Table I
ETL, echo train length; IQ, image quality; FS, fat suppression; Red, 6; Green, 10.
Femur Tibia Patella Femur Tibia Patella Femur Tibia Patella
T1 3DSPGR +FS (WE)
1.5 mm sections, 16 cm
FOV, 512x512 to 245x256
Artifacts
Sagittal
Coverage = 10
IQ = 8 (poor at med/lat
margins)
Coverage = 10
IQ = 8 (poor at tibial
spines)
Coverage = 10
IQ = 7 (surfaces oblique)
Coverage = 10
IQ = 8 (poor at med/lat
margins; no T2)
Coverage = 10
IQ = 8 (poor at tibial
spines; no T2)
Coverage = 10
IQ = 8 (surfaces oblique;
no T2)
Coverage = 5 (excludes
med and lat margins)
IQ = 9
Coverage = 5 (excludes
med and lat margins)
IQ = 9
Coverage = 5 (excludes
med and lat margins)
IQ = 9
Coronal
Coverage = 5 (excludes
ant and post surfaces)
IQ = 10 (excellent for
central surface)
Coverage = 10
IQ = 9 (poor at ant/post
margins) Coverage = 0
Coverage = 5 (excludes
ant and post surfaces)
IQ = 8 (excellent for
central surface; no T2)
Coverage = 10
IQ = 8 (no T2) Coverage = 0
Coverage = 5 (excludes
ant and post margins)
IQ = 9
Coverage = 5 (excludes
ant and post margins)
IQ = 9 Coverage = 0
Axial Coverage = 0 Coverage = 0
Coverage = 10
IQ = 10 Coverage = 0 Coverage = 0
Coverage = 10
IQ = 9 (no T2)
Coverage = 3 (captures
part of ant and post
margins)
IQ = 9 Coverage = 0
Coverage = 5 (excludes
sup and inf margins)
IQ = 9
Sagittal + coronal
Coverage = 10
IQ = 10 (max central
surface)
Coverage = 10
IQ = 10
Coverage = 10
IQ = 7 (surfaces oblique)
Coverage = 10
IQ = 9 (no T2)
Coverage = 10
IQ = 9 (no T2)
Coverage = 10
IQ = 8 (surfaces oblique;
no T2)
Coverage = 10
IQ = 9
Coverage = 10
IQ = 9
Coverage = 5 (excludes
med and lat margins)
IQ = 9
Coronal + axial
Coverage = 5 (excludes
ant and post surfaces)
IQ = 10 (max central
surface)
Coverage = 10
IQ = 9 (poor at ant/post
margins)
Coverage = 10
IQ = 10
Coverage = 5 (excludes
ant and post surfaces)
IQ = 8 (excellent for
central surface; no T2)
Coverage = 10
IQ = 8 (no T2)
Coverage = 10
IQ = 9 (no T2)
Coverage = 8
IQ = 9
Coverage = 5 (excludes
ant and post margins)
IQ = 9
Coverage = 5 (excludes
sup and inf margins)
IQ = 9
Sagittal + axial
Coverage = 10
IQ = 8 (poor at med/lat
margins)
Coverage = 10
IQ = 8 (poor at tibial
spines)
Coverage = 10
IQ = 10
Coverage = 10
IQ = 8 (poor at med/lat
margins; no T2)
Coverage = 10
IQ = 8 (poor at tibial
spines; no T2)
Coverage = 10
IQ = 9 (no T2)
Coverage = 6 (better ant
/ post, but no med / lat)
IQ = 9
Coverage = 5 (excludes
med and lat margins)
IQ = 9
Coverage = 10
IQ = 9
Sagittal+coronal+axial
Coverage = 10
IQ = 10 (max central
surface)
Coverage = 10
IQ = 10
Coverage = 10
IQ = 10
Coverage = 10
IQ = 9 (no T2)
Coverage = 10
IQ = 9 (no T2)
Coverage = 10
IQ = 9 (no T2)
Coverage = 10
IQ = 9
Coverage = 10
IQ = 9
Coverage = 10
IQ = 9
IW (30-40 ms) FSE + FS
(WE)
ETL 8, 3/0 mm, 16 cm,
256x256
Sagittal N/A N/A N/A
Coverage = 10
IQ = 8 (poor at med/lat
margins)
Coverage = 10
IQ = 8
(poor at tibial
spines)
Coverage = 10
IQ = 7 (surfaces oblique)
Coverage = 5 (excludes
med and lat margins)
IQ = 9
Coverage = 5 (excludes
med and lat margins)
IQ = 9
Coverage = 5 (excludes
med and lat margins)
IQ = 9
Coronal N/A N/A N/A
Coverage = 5 (excludes
ant and post surfaces)
IQ = 8 (excellent for
central surface)
Coverage = 10
IQ = 8 Coverage = 0
Coverage = 5 (excludes
ant and post margins)
IQ = 9
Coverage = 5 (excludes
ant and post margins)
IQ = 9 Coverage = 0
Axial N/A N/A N/A Coverage = 0 Coverage = 0
Coverage = 10
IQ = 8
Coverage = 3 (captures
ant and post margins)
IQ = 9 Coverage = 0
Coverage = 5 (excludes
sup and inf margins)
IQ = 9
Sagittal + coronal N/A N/A N/A
Coverage = 10
IQ = 9
Coverage = 10
IQ = 8
Coverage = 10
IQ = 7 (surfaces oblique)
Coverage = 10
IQ = 9
Coverage = 10
IQ = 9
Coverage = 5 (excludes
med and lat margins)
IQ = 9
Coronal + axial N/A N/A N/A
Coverage = 5 (excludes
ant and post surfaces)
IQ = 8 (excellent for
central surface;)
Coverage = 10
IQ = 8
Coverage = 10
IQ = 8
Coverage = 8
IQ = 9
Coverage = 5 (excludes
ant and post margins)
IQ = 9
Coverage = 5 (excludes
sup and inf margins)
IQ = 9
Sagittal + axial N/A N/A N/A
Coverage = 10
IQ = 8 (poor at med/lat
margins)
Coverage = 10
IQ = 8 (poor at tibial
spines)
Coverage = 10
IQ = 8
Coverage = 6 (better ant
/ post, but no med / lat)
IQ = 9
Coverage = 5 (excludes
med and lat margins)
IQ = 9
Coverage = 10
IQ = 9
Sagittal+coronal+axial N/A N/A N/A
Coverage = 10
IQ = 9
Coverage = 10
IQ = 8
Coverage = 10
IQ = 8
Coverage = 10
IQ = 9
Coverage = 10
IQ = 9
Coverage = 10
IQ = 9
Osteophytes
Cartilage
score
Cartilage
volume/thickness
A101
Osteoarthritis and Cartilage Vol. 14, Supplement A
Fig. 6. Reproducible alignment of coronal plane. (A) Coronal sections aligned tangentially to the posterior margins of th e posterior femoral
condyles on axial localizer, with knee positioned acquired with the patella centered in coil. (B) To adjust for variations in knee flexion during
imaging, coronal sections should be secondarily aligned on a sagittal localizer parallel to the femoral diaphysis or perpendicular to the tibial
plateau, depending on which cartilage surface is the primary focus of study. (C) Posterior femoral condyles appear symmetrically positioned
on properly aligned coronal image.
Fig. 7. Reproducible alignment of sagittal plane. (A) Sagittal sections aligned perpendicularly to a line tangential to the posterior margins of the
posterior femoral condyles on axial localizer acquired with knee positioned with the patella centered in coil. (B) Sagittal sections are secondary
aligned parallel to femoral diaphysis on coronal localizer. This is often not critical because the knee coil constrains abductioneadduction of leg.
(C) Properly aligned sagittal image.
Fig. 8. Reproducible alignment of axial plane. (A) Axial sections aligned perpendicularly to the articular surface of the patella on sagittal local-
izer acquired with knee positioned with the patella centered in coil. Note that coverage in this example is restricted to the patella and does not
include the femorotibial compartments. (B) Axial sections are secondary aligned perpendicular to a line parallel to the femoral diaphysis on the
coronal localizer. (C) Properly aligned axial image.
A102
C. G. Peterfy et al.: MRI protocols for whole-organ assessment
on coronal images. Accordingly, both planes are usually
needed for thorough assessment.
Ligaments and tendons are best examined with long-TE
MRI techniques because the so-called ‘‘magic-angle’’
effects
15
on short-TE images can produce foci of high signal
intensity within these structures, mimicking inflammation
and tear. Indeed, the magic-angle phenomenon must be
considered in all collagen-containing structures, including
Fig. 9. Cartilage contrast with various pulse sequences. (A) Sagittal fat-suppressed T1-weighted 3D GRE image depicting articular cartilage as
a high-signal structure in sharp contrast against adjacent low-signal bone, marrow fat, intra-articular adipose, fluid, ligaments and menisci. (B)
Sagittal 3D DESS image showing partial-thickness cartilage defect (arrow) over posterior lateral tibial. Note the similarities in contrast prop-
erties of fat-suppressed DESS with those of fat-suppressed FSE. (C) Sagittal fat-suppressed IW 2D FSE shows a loose body (arrow) in the
patellofemoral compartment. (D) Sagittal T2-weighted 2D FSE image without FS shows a partial-thickness defect (arrow) of the lateral femoral
cartilage adjacent to the posterior horn of the meniscus. (E) Sagittal fat-suppres sed T2-weighted 2D FSE image of a different knee shows
a partial-thickness cartilage defect (arrow) in a similar location.
Fig. 10. T1-weighted SE offers poor cartilageefluid contrast. Sagittal T1-weighted SE image (A) shows poor contrast between articular car-
tilage and adjacent joint fluid compared to that with T2-weighted FSE (B). Both techniques delineate osteophytes well. Note the pronounced
bone attrition of the lateral tibial plateau.
A103
Osteoarthritis and Cartilage Vol. 14, Supplement A
articular cartilage
16
and the menisci
17
(Fig. 18). Neverthe-
less, IW FSE images are usually adequate. As for the
menisci, fat-suppressed SSFP techniques show promise
for assessing the cruciate and collateral ligaments
6
, but
experience in this respect is still limited. The anterior cruci-
ate and posterior cruciate ligaments are typically assessed
in the sagittal plane (Fig. 19), whereas the medial collateral
ligament and lateral collateral ligament are best assessed in
the coronal plane (Fig. 20).
MRI is also useful for assessing synovitis and joint effu-
sions. Some degree of synovial thickening can be found
in a majority of osteoarthritic joints
18
. Whether this synovitis
contributes directly to articular cartilage loss in OA, or sim-
ply arises in reaction to the breakdown of cartilage by other
causes remains a controversy
19
. However, synovitis may
be important to the symptoms and disability of OA, and
may pose different treatment requirements than cartilage
loss does. Synovial effusions are easily depicted with con-
ventional fat-suppressed T2-weighted or IW FSE. However,
large effusions can distend the suprapatellar recess beyond
the superior limit of the FOV (Fig. 21), making accurate
quantification of effusion volume impossible. It may be nec-
essary in such cases to extend the FOV superiorly, making
sure to retain coverage of the tibial cartilage and any bone
marrow abnormality in the tibial plateau. Another challenge
to quantifying synovial effusion or synovial tissue is accu-
rately differentiating these structures from each other. This
can sometimes be difficult with conventional fat-suppressed
T2-weighted or IW FSE, and special techniques may be
required (Figs. 22 and 23), such as magnetization-transfer
subtraction; specifically windowed fat-suppressed, T1-
weighted imaging; or intravenous injection of gadolinium
(Gd)-containing contrast material
20e22
. Gd-containing con-
trast can also be used to quantify the severity of the syno-
vitis by measuring the rate of synovial enhancement over
time using rapid, sequential MRI. The majority of work in
this area has, however, focused on rheumatoid arthritis
thus far, and experience with OA is still limited. Additionally,
diffusion of contrast material from synovial tissue into adja-
cent synovial fluid rapidly obscures the boundary between
Fig. 11. Discriminating cartilageecartilage interfaces can be difficult. (A) Example of a coronal fat-suppressed T1-weighted 3D GRE image
showing po or discrimination between abutting articular cartilage plates in the media l femorotibial compartment, and between th e lateral
meniscus and adjacen t s ynovial fluid. This is not always the case with 3D GRE (see Figs.2,3and9). (B) 3D DESS image of the same
knee discriminates the cartilageecartilage and meniscusefluid interfaces.
Fig. 12. Osteophyte imaging. (A) Sagittal 3D GRE image shows well-delineated marginal osteophytes (large arrows) of the superior and
inferior patella, the anterior and posterior femur and the anterior tibia. (B) Sagittal 3D DESS image shows several subchondral osteophytes
(small arrows) of the patellofemoral compartment, as well as an osteophyte of the superior patellar margin.
A104
C. G. Peterfy et al.: MRI protocols for whole-organ assessment
Fig. 13. Failed FS mimics bone marrow edema. (A) Sagittal FS IW FSE image shows subarticular cysts and marrow edema in the lateral tibial
plateau. High signal within the patella is also consistent with bone marrow edema, but the association of high signal extending across multiple
adjacent tissue compartments (prepatellar soft tissues, intra-articular fat pads, anterior femur) suggests failed FS as an alternative cause. This
reduces the diagnostic accuracy of this scan for patellar bone marrow edema. (B) Sagittal FS IW FSE image of a different knee shows bone
marrow edema in the anterior lateral femur beneath a focus of high signal in the articular cartilage indicative of chondromalacia. The ill-defined
margins of this focus of marrow edema make it difficult to segment ac curately for volumetric quantification. The patella shows diffuse hig h
signal in the marrow associated with subarticular cysts beneath a focal full-thickness cartilage defect. High signal in the superior prepatellar
tissues (asterisks) shows a pattern consistent with failed FS, but the pattern in the inferior prepatellar tissues is more consistent with infrapa-
tellar bursitis. This suggests that the FS failure is localized and raises the specificity of the diagnosis of marrow edema in the patella.
Fig. 14. Metallic artifact. (A) Sagittal IW FSE image shows severe metallic artifacts in femur and tibia associated with surgical anterior cruciate
ligament repair. (B) Fat-suppressed T1-weighted 3D GRE image of the same knee shows worse artifacts with pronounced failure of FS. Sag-
ittal IW FSE (C) and fat-suppressed, T1-weighted GRE (D) images show a mild artifact adjacent to the anterior femoral cartilage associated
with a tiny metal particle from previous arthroscopic procedure. The artifact (arrow) obscures slightly the cartilage surface on the GRE image
(D), but is barely perceptible on the FSE image (C).
A105
Osteoarthritis and Cartilage Vol. 14, Supplement A
these two components, and complicates the practical appli-
cation of this approach
20,23
(Fig. 22).
Finally, periarticular bursitis and cysts can be causes of
pain and dysfunction in the knee that mimic or contribute
to the clinical presentation of OA
11,24,25
. The most com-
mon of these are popliteal cysts, which communicate
with the synovial cavity through a space between the
semimembranosis and medial gastocnemius tendons.
These cysts are easily diagnosed in both the axial and
sagittal planes (Fig. 24) using conventional fat-sup-
pressed T2-weighted or IW FSE. Most popliteal cysts
are small and clinically occult, but some grow to several
hundred ml and dissect down the calf and/or up the thigh.
These large cysts can cause local swelling and interfere
with knee flexion. They also occasionally rupture causing
local inflammation and acute calf pain. In some cases, it
may be necessary to shift the centering of the FOV
caudally in order to completely include a large dissecting
popliteal cyst. Care must be taken when doing this not to
exclude other important structures, such as the suprapa-
tellar recess. Cysts of the proximal tibiofibular joint, anser-
ine bursitis, semimembranosis bursitis and prepatellar
bursitis are additional potential causes of pain and
swelling about the knee that are sensitively identified
and easily discriminated from one another with conven-
tional fat-suppressed T2-weighted or IW FSE. Failed FS
can obscure or mimic prepatellar bursitis.
These considerations are summarized in Table I for two
of the MRI techniques discussed above, T1-weighted 3D
SPGR with FS or WE, and IW 2D FSE with FS or WE, using
imaging parameters typical for 1.5 T MRI. The table grades
each of these techniques for anatomical coverage and IQ
(for those regions adequately covered) with respect to
quantifying or scoring articular cartilage morphology,
Fig. 15. Aliasing artifact. (A) Sagittal GRE image shows aliasing, or ‘‘wrap’’ artifact, obscuring the anterior portion of the patella but not the
articular cartilage. This type of artifact superimposes anatomy from outside of the FOV over anatomy within the FOV. (B) Sagittal GRE image
with wrap artifact obscuring the central femorotibial cartilage. The trained eye can often cope with mild wrap artifacts, but this can be a signif-
icant problem for automated image-processing or analysis software.
Fig. 16. Pulsation artifact. (A) Mid-sagittal fat-suppressed IW 2D FSE image shows pronounced pulsation artifacts from the popliteal artery
along the phase-encoding directi on (anterioreposterior in this case) obscuring the articular anatomy of the knee. Note also, the failed FS
at the superior and inferior ends of the image due to field heterogeneity in these locations. (B) Axial 3D DESS image shows mild pulsation
artifacts (large arrow) from the popliteal artery. These artifacts do not obsc ure any articular anatomy in this case because phase encoding
was ordered medialelateral. Note also the partial-thickness cartilage defect over the medial facet of the patella (small arrow).
A106
C. G. Peterfy et al.: MRI protocols for whole-organ assessment
osteophytes, bone marrow abnormality, subarticular bone
cysts, menisci, ligaments and synovial effusion. Anatomical
coverage and IQ were each graded from 0 to 10 (0 ¼ not
useable, .,10¼ ideal) based on consensus by the panel
of experts described above.
Other key constraints in any imaging protocol are patient
tolerance and safety, and the availability of requisite
hardware and software. Patient tolerance depends on the
temperament and physical condition of the patient, as well
as the duration of the protocol and its individual pulse se-
quences. While fewer than 5% of subjects actually cannot
complete an MRI examination because of claustrophobia,
many find the experience unpleasant, and except under
special circumstances, extending the examination time be-
yond 75 min carries significant risk of subject drop-out. Ide-
ally, the examination should be kept under 60 min.
Additionally, the longer the individual pulse sequence is in
a protocol, the greater the risk of subject motion artifacts
on the images. This is most problematic with 3D se-
quences, such as those used to quantify articular cartilage.
Depending on the spatial resolution required, these scans
can take longer than 12 min to acquire. A number of recent
innovations, such as frequency-selective water-excitation
and SSFP techniques, however, may reduce these acquisi-
tion times considerably
5
.
Fig. 17. GRE techniques are insensitive for bone marrow edema, but excellent for delineating cysts. Fat-suppressed 3D GRE image (upper left
panel) shows a small focal cartilage defect (arrow) over the posterior femoral condyle, but does not disclose the subjacent marrow edema which
is visible on the fat-suppressed T2-weighted FSE image (lower left panel). Images acquired 22 weeks later (right panels) show that a subarticular
cyst has developed in this location. Fat-suppressed T2-weighted FSE image (lower right panel) shows marrow edema adjacent to this cyst.
Fig. 1 8. Mag ic-an gle phe nomen on in the meniscu s. Coron al fat-
suppressed T1-weighted 3D GRE image shows high signal (arrow)
in the medial segment of the posterior horn of the lateral meniscus.
Annular collagen fibers in this region of the meniscus are oriented
at the critical angle of 55( relative to the static magnetic field (B
0
)
and therefore exhibit prolonged T2 relaxation, or magic-angle phe-
nomenon. Thi s phenomenon can affect any collagen-containing
structure, including articular cartilage and ligaments, and can mimic
tissue damage or reduce contrast with adjacent structures.
A107
Osteoarthritis and Cartilage Vol. 14, Supplement A
Patient safety considerations relate primarily to effects of
high magnetic fields on metal or magnetically sensitive de-
vices in the subject’s body or in the room near the mag-
net
26
. Accordingly, whole-body MRI is contraindicated in
patients with aneurysm clips, pacemakers, cochlear im-
plants, metallic splinters in the eye from metal-grinding in-
juries, and a host of other surgical and accidental
conditions. Recently introduced low-field strength extremity
MRI systems obviate some of these biohazards when the
vulnerable parts of the body can be kept out of the bore
of the magnet and away from the main magnetic field
27e29
.
These extremity systems as well as open magnet designs
also improve patient comfort. The lower field strength,
however, generally limits spatial resolution and the effec-
tiveness of spectral FS or WE. Nevertheless, these systems
may be applicable to some experimental designs or logisti-
cal circumstances.
IQ control and data management
Once the imaging sites have been selected, the imaging
protocol designed and the technologists trained, image ac-
quisition, transfer and quality must be closely supervised
to ensure a high-quality image set for analysis. Variability
in IQ can be controlled by preparing the subject for the ex-
amination and by proper calibration and maintenance of
the imaging system. Device performance is documented
and maintained by performing device quality control. One
aspect of this is done by the imaging sites as part of their
routine clinical quality control, but additional study-specific
quality control often must also be performed. As men-
tioned above, this may require the use of specialized
phantoms.
After the images are acquired according to the study-spe-
cific protocol, they must be reviewed for protocol compli-
ance, patient positioning, anatomical coverage, and IQ,
including any artifacts
30
. In addition to FS failure, chemi-
cal-shift, motion, partial-volume averaging, and magic-angle
phenomenon, which were discussed above, other common
artifacts include metallic artifact; aliasing, or ‘‘wrap’’; vascu-
lar pulsation (a special case of motion); and truncation.
Metallic artifacts are commonly encountered in knees that
contain hardware from prior surgery, such as cruciate liga-
ment repair or internal fixation for fracture. Even in the ab-
sence of actual hardware, micrometallic particles from
instrumentation during arthroscopy or open arthrotomy
can create small problematic artifacts in some cases
Fig. 19. Cruciate ligament contrast. Sagittal fat-suppressed T1-weighted 3D GRE (A) offers poor contrast for the anterior cruciate ligament.
Fat-suppressed T2-weighted 2D FSE (B) and 3D DESS (C), however, delineate the cruciate ligaments well.
Fig. 20. Collateral ligament contrast. (A) Coronal fat-suppressed IW FSE shows partial tear of the medial collateral ligament (small arrow), with
delamination of the inner and outer leaves. (B) Coronal DESS image of a different knee shows good delineation of the medial collateral lig-
ament (small arrow) but high signal (large arrow) in lateral tibial cartilage indicative of collagen-matrix degeneration.
A108
C. G. Peterfy et al.: MRI protocols for whole-organ assessment
(Fig. 14). GRE techniques, such as 3D SPGR are more
sensitive to metallic artifacts than are SE techniques, partic-
ularly FSE. Aliasing or wrap artifact presents as anatomy
from outside of the FOV superimposed on anatomy within
the FOV (Fig. 15). The trained eye can often cope with
mild wrap artifacts, but this can be a significant problem
for automated image-processing or image-analysis soft-
ware. Random patient motion during image acquisition
can cause blurring and ghosting artifacts along the phase-
encoding direction. This problem is amplified with 3D imag-
ing, as phase encoding is applied in two planes rather than
one, as in 2D imaging. In some cases, patient movement
can be ameliorated by verbally calming or medically sedat-
ing the patient, but patients with Parkinson’s disease or
other causes of involuntary tremor can be extremely chal-
lenging to image. A more common cause of involuntary
motion is vascular pulsation (Fig. 16). In the knee, pulsation
artifacts from the popliteal artery often degrade mid-sagittal
anatomy. This can often be improved by applying saturation
bands outside of the FOV to suppress signal in moving
blood and therefore in the pulsation artifacts. Additionally,
the direction of these artifacts and therefore the anatomy
that they potentially obscure can be controlled by adjusting
the phase-encoding direction. Truncation artifacts occur
when the number of phase-encoding steps of high spatial
frequencies is insufficient for accurate reproduction of ana-
tomic detail. These artifacts have been blamed for the ap-
pearance of lamina within articular cartilage
31
; although,
this is still controversial
32
.
If the images are of acceptable quality based on explicit
IQ criteria, they are entered into the study database. Stud-
ies that involve multiple imaging sites typically use a central
database, as a central reading radiologist is required to
limit variability in the image analysis and interpretation.
All processes performed by the central radiology service
must be done in strict accordance with Standard Operating
Procedures (SOPs) and Study-Specific Procedures
(SSPs). If the image data do not pass the incoming quality
inspection, a decision needs to be made as to whether or
not the data can be corrected, for example by using infor-
mation obtained from the instrument quality control. If the
images cannot be salvaged, the imaging must be repeated
or the data point discarded. The criteria applied in this are
again controlled by SOPs, and data exclusions must be
carefully documented. If repeat imaging is necessary
a feedback loop must be designed with appropriate timing
criteria. Final data consistency checks are applied before
the data are submitted for filing, and there may be an op-
portunity for a final adjustment if supported by appropriate
instrument quality control information. The ultimate result of
this process is a high-quality image set that conforms to
rigorous quality assurance principles and can support
high-quality image analysis for whole-organ assessment
Fig. 22. Gd-enhanced delineation of synovium. Sequential sagittal
T1-weighted images through the s uprapatellar recess of a knee
with rheumatoid arthritis were acquired before and 5 min, 15 min
and 4 5 min following bolus i.v. injection of Gd-containing contrast
material. Five minutes following injection, high-signal synovial tis-
sue can be differentiated from adjacent low-signal joint fluid. How-
ever, with time, the contrast material leaks out of the synovium into
the joint fluid and obscures the synoviumefluid interface.
Fig. 21. Large effusions can extend beyond the FOV. Sagittal fat-
suppressed IW FSE image shows a distended suprapatellar recess
extending beyond the cranial limit of the FOV (arrow). In these
cases, the FOV should be centered more cranially, while still includ-
ing at least the entire tibial cartilage.
Fig. 23. Synovial imaging without Gd-containing contrast. Axial fat-suppressed T1-weighted 3D SPGR (A) and magnetization-transfer subtrac-
tion GRE (B) images at the level of the suprapatellar recess show good discrimination between synovial tissue and joint fluid.
A109
Osteoarthritis and Cartilage Vol. 14, Supplement A
of the knee in OA clinical trials and epidemiological
studies.
References
1. Peterfy C, Guermazi A, Zaim S, Tirman P, Miaux Y,
White D, et al. Whole-Organ Magnetic Resonance Im-
aging Score (WORMS) of the knee in osteoarthritis.
Osteoarthritis Cartilage 2004;12(3):177e90.
2. Flanders AE, Carrino JA. Understanding DICOM and
IHE. Semin Roentgenol 2003;38(3):270e81.
3. Kneeland JB, Hyde JS. High-resolution MR imaging
with local coils. Radiology 1989;171:1e7.
4. Mosher T, Dardzinski B, Smith M. Human articular
cartilage: influence of aging and early symptomatic de-
generation on the spatial variation of T2 e preliminary
findings at 3 T. Radiology 2000;241(1):259e66.
5. Gold GE, Vasanawala SS, Pauly JM, Nishimura DG.
MR Imaging of Knee Cartilage with FEMR. In: Interna-
tional Skeletal Society, September 1e8, 2001, Que-
bec City, 2001.
6. Kreitner KF, Hansen M, Schadmand-Fischer S,
Krummenauer F, Runkel M. Low-field MRI of the
knee joint: results of a prospective, arthroscopically
controlled study. Rofo Fortschr Geb Rontgenstr Neuen
Bildgeb Verfahr 1999;170(1):35e40.
7. Ruehm S, Zanetti M, Romero J, Hodler J. MRI of patel-
lar articular cartilage: evaluation of an optimized gradi-
ent echo sequence (3D-DESS). J Magn Reson
Imaging 1998;8(6):1246e51.
8. Mosher TJ, Pruett SW. Magnetic resonance imaging
of superficial cartilage lesions: role of contrast in
lesion detection. J Magn Reson Imaging 1999;10(2):
178e82.
9. Eckstein F, Sittek H, Milz S, Putz R, Reiser M. The mor-
phology of articular cartilage assessed by magnetic
resonance imaging (MRI). Reproducibility and ana-
tomical correlation. Surg Radiol Anat 1994;16(4):
429e38.
10. Nishii T, Nakanishi K, Sugano N, Masuhara K,
Ohzono K, Ochi T. Articular cartilage evaluation in
osteoarthritis of the hip with MR imaging under contin-
uous leg traction. Magn Reson Imaging 1998;16:
871e5.
11. Peterfy CG. Medial knee pain in the aging patient. In:
Klippel J, Dieppe P, Eds. Rheumatology 2E. Philadel-
phia: Mosby 1998:8.12.8e8.12.9.
12. Felson DT, Chaisson CE, Hill CL, Totterman SM,
Gale ME, Skinner KM, et al. The association of bone
marrow lesions with pain in knee osteoarthritis. Ann In-
tern Med 2001;134(7):541e9.
13. Felson D, McLaughlin S, Goggins J, LaValley M,
Gale ME, Totterman S, et al . Bone marrow edema
(BME) and its relation to X-ray progression in knee os-
teoarthritis (OA). Arthritis Rheum 2003;46(9):S558.
14. Beltran J, Shankman S. The menisci. In: Davies AM,
Cassa-Pullicino VN, Eds. Imaging of the Knee. New
York: Springer-Verlag 2002:129e152.
15. Erickson SJ, Prost RW, Timins ME. The ‘‘magic angle’’
effect: background physics and clinical relevance.
Radiology 1993;188:23e5.
16. Rubenstein JD, Kim JK, Morava-Protzner I, Stanchev PL,
Henkelamn RM. Effects of collagen orientation on MR
imaging characteristics of bovine cartilage. Radiology
1993;188:219e26.
17. Peterfy CG, Janzen DL, Tirman PFJ, van Dijke CF,
Pollack M, Genant HK. Magic-angle phenomenon:
a cause of increased signal in the normal lateral me-
niscus on short-TE MR images of the knee. AJR Am
J Roentgenol 1994;163:149e54.
18. Fernandez-Madrid F, Karvonen RL, Teitge RA,
Miller PR, An T, Negendank WG. Synovial thickening
detected by MR imaging in osteoarthritis of the knee
confirmed by biopsy as synovitis. Magn Reson Imag-
ing 1995;13:177e83.
19. Brandt KD. Insights into the natural history of osteoar-
thritis and the potential for pharmacologic modification
of the disease afforded by study of the cruciate-defi-
cient dog. In: Keutner KE, Goldberg VM, Eds. Osteo-
arthritic Disorders. Rosemont, IL: American Academy
of Orthopaedic Surgeons 1995:419e426.
20. Peterfy CG, Majumdar S, Lang P, van Dijke CF,
Sack K, Genant H. MR imaging of the arthritic knee:
improved discrimination of cartilage, synovium and
effusion with pulsed saturation transfer and fat-
suppressed T1-weighted sequences. Radiology 1994;
191:413e9.
21. Yamato M, Tamai K, Yamaguchi T, Ohno W. MRI
of the knee in rheumatoid arthritis: Gd-DTPA perfu-
sion dynamics. J Comput Assist Tomogr 1993;17:
781e5.
22. Ko
¨
nig H, Sieper J, Wolf KJ. Rheumatoid arthritis: eval-
uation of hypervascular and fibrous pannus with dy-
namic MR imaging enhanced with Gd-DTPA.
Radiology 1990;176:473e7.
23. Drape
´
J-L, Thelen P, Gay-Depassier P, Silbermann O,
Benacerraf R. Intraarticular diffusion of Gd-DOTA after
intravenous injection in the knee: MR imaging evalua-
tion. Radiology 1993;188:227e34.
24. Janzen DL, Peterfy CG, Tirman PFJ, Genant HK. Cys-
tic lesions around the knee joint: MR imaging findings.
AJR Am J Roentgenol 1994;163:155e61.
25. Morrison JL, Kaplan PA. Water on the knee: cysts, bur-
sae, and recesses. Magn Reson Imaging Clin N Am
2000;8(22):349e70.
26. Shellock FG, Crues JV. MR procedures: biologic
effects, safety, and patient care. Radiology 2004;
232(3):635e52. Epub 2004 Jul 29.
Fig. 24. Imaging popliteal cysts. Medial sagittal fat-suppressed IW
FSE image shows a small popliteal cyst (asterisk) between the ten-
dons of the semimembranosis and medial gastrocnemius (arrow).
A110
C. G. Peterfy et al.: MRI protocols for whole-organ assessment
27. Shellock FG, O’Neil M, Ivans V, Kelly D, O’Connor M,
Toay L, et al. Cardiac pacemakers and implantable
cardioverter defibrillators are unaffected by operation
of an extremity MR imaging system. AJR Am J Roent-
genol 1999;172(1):165e70.
28. Shellock FG, Crues JV. Aneurysm clips: assessment of
magnetic field interaction associated with a 0.2-T ex-
tremity MR system. Radiology 1998;208(2):407e9.
29. Peterfy CG, Roberts T, Genant HK. Dedicated extrem-
ity MRI: an emerging technology. In: Kneeland JB, Ed.
Radiol Clin N Am. Philadelphia, PA: W.B. Saunders
1996:1e20.
30. Peh WC, Chan JH. Artifacts in musculoskeletal mag-
netic resonance imaging: identification and correction.
Skeletal Radiol 2001;30(4):179e91.
31. Frank LR, Brossmann J, Buxton RB, Resnick D. MR
imaging truncation artifacts can create a false laminar
appearance in cartilage. AJR Am J Roentgenol 1997;
168(2):547e54.
32. Kim DJ, Suh JS, Jeong EK, Shin KH, Yang WI. Corre-
lation of laminated MR appearance of articular carti-
lage with histology, ascertained by artificial
landmarks on the cartilage. J Magn Reson Imaging
1999;10(1):57e64.
Syllabus of abbreviations
2D: two-dimensional.
3D: three-dimensional.
DESS: dual-echo steady-state.
DICOM: Digital Imaging and Communications in Medicine.
FLASH: fast low-angled shot.
FOV: field of view.
FSE: fast spin-echo.
Gd: gadolinium.
GRE: gradient-echo.
IW: intermediate-weighted.
NEX: number of excitations.
OA: osteoarthritis.
OMERACT: Outcome Measures in Rheumatology Clinical Trials.
OARSI: Osteoarthritis Research Society International.
SOP: Standard Operating Procedure.
S/N: signal-to-noise ratio.
SPGR: spoiled gradient recalled.
SSFP: steady-state free precession.
SSP: Study-Specific Procedure.
STIR: short-tau inversion recovery.
TE: echo time.
TR: repetition time.
A111
Osteoarthritis and Cartilage Vol. 14, Supplement A
    • Furthermore, T 1w GE sequences can adopt Fat Suppression (FS) to increase the contrast between the bone and cartilage removing most of the signal from the bone, so providing a sharper delineation of the BCI[5]. Three-dimensional fat suppressed spoiled gradient-echo (FS SPGR)[5]and DESS (Dual Echo Steady State)[17]sequences are commonly used in clinical examinations for the quantitative imaging of the cartilage. SPGR[7]and DESS sequences adopt FS to increase the contrast of the cartilage and evidence micro lesions.
    [Show abstract] [Hide abstract] ABSTRACT: Segmentation of cartilage from Magnetic Resonance (MR) images has evolved as a tool for the diagnosis of knee joint pathologies. However, accuracy and reproducibility of automated methods of cartilage segmentation may require the prior extraction of bone surfaces from MR imaging sequences specifically designed to evidence the cartilage and not the bone. Thus a priori knowledge of knee joint structures and fully automated segmentation methods are adopted to provide reliable detection of bone surfaces. In this paper, we review knee bone segmentation methods from MR images. We classified the methods proposed in literature according to the level of a priori knowledge, the level of automation and the level of manual user interaction. Furthermore we discuss the segmentation results in literature in relation to the MR sequences used to image the bone.
    Full-text · Article · Feb 2016
    • Subchondral bone marrow lesions (BMLs) are common findings on magnetic resonance (MR) images of knees with osteoarthritis (OA) and relate to structural and symptomatic progression of OA123. While BMLs are often assessed on intermediate-weighted fat suppressed (IW FS) or similar sequences [4] some researchers have also evaluated BMLs on 3-dimensional dual echo steady state (3D DESS) sequences or other similar sequences that are used for cartilage mea- surements [5] . The latter approach enables a time-and costefficient method to assess changes in BMLs and cartilage on the same sequence [6] .
    [Show abstract] [Hide abstract] ABSTRACT: Subchondral bone marrow lesions (BMLs) are related to structural and symptomatic osteoarthritis progression. However, it is unclear how sequence selection influences a quantitative BML measurement and its construct validity. We compared quantitative assessment of BMLs on intermediate-weighted fat suppressed (IW FS) turbo spin echo and 3-dimensional dual echo steady state (3D DESS) sequences. We used a customized software to measure 30 knees’ (24- and 48-month MR images) BMLs on both sequences. The results showed that the IW FS sequences have much larger BML volumes (median: IW FS = 1840 mm 3 ; DESS = 191 mm 3 ) and BML volume change (between 24 and 48 months) than DESS sequence and demonstrate more BML volume change. The 24-month BML volume on IW FS is correlated with BML volume on DESS ( r s = 0.83). BML volume change on IW FS is not significantly correlated with change on DESS. The 24-month WOMAC pain is correlated with the 24-month BMLs on IW FS ( r s = 0.39) but not DESS. The change in WOMAC pain is correlated with BML volume change on IW FS ( r s = 0.37) but not DESS. Overall, BML quantification on IW FS offers better validity and statistical power than BML quantification on a 3D DESS sequence.
    Full-text · Article · Nov 2015
    • Multi echo sequence with spin echo sequence was used for T2 mapping, and the image parameters were as follows: repetition time = 1500 ms, echo time = 13, 26, 39, 52, 65, 78, 91 and 104 ms, field of view = 150 × 150 mm, slice thickness 3.0 mm, matrix 256 × 256, band width 233 Hz/pixel. The coronal plane[21,22]is excellent for evaluating articular cartilage along the central weight-bearing surfaces of the femur and tibia. The present study used the medial femoral condyle and tibia on the coronal plane to compare the KAM and MCF during gait.
    Article · Jan 2015
    • The coronal proton density-weighted sequence used in this study lacks the signal-to-noise ratio and spatial resolution required to demonstrate the same degree of subchondral trabecular detail compared to dedicated trabecular imaging sequences such as 3D gradient echo imaging [16, 17]. However, the utility of the coronal proton density-weighted sequences to image the osteochondral unit has been previously described [30] as well as the ability to depict subchondral bone changes in OA such as early osteophyte formation [31]. The contrast resolution of the coronal proton density weighted images also allows good depiction of areas of low-signal sclerosis against a background of high signal bone marrow (Fig. 2).
    [Show abstract] [Hide abstract] ABSTRACT: Objective: To determine whether differences in subchondral sclerosis at the tibial plateau could be detected with magnetic resonance (MR) imaging in two different age groups. Materials and methods: This was a retrospective hypothesis-testing study. Thirty-two knees in group A (25-30 year olds) and 32 knees in group B (45-50 years old) were included. Participants had no MR features of osteoarthritis (OA). On coronal images, tibial articular cartilage thickness was measured, and regions of interest were created in the medial and lateral tibial plateau subchondral bone and in the tibial metaphysis. The measure of heterogeneity at the tibial plateaux was the ratio of the standard deviation of the signal in the medial/lateral compartment to the standard deviation of the signal in the metaphysis (ratio of standard deviations--RSS(medial)/RSS(lateral)). Differences between groups were assessed using unpaired Student's t-tests. Results: Mean RSS(medial) was 2.61 (standard deviation, SD = 0.77) in group A and 2.97 (SD = 0.59) in group B. Mean RSS(lateral) in group A was 1.86 (SD = 0.63) and 1.89 (SD = 0.43) in group B. Mean total cartilage thickness (in mm) in group A was 3.38 (SD = 0.90) for the medial and 3.90 (SD = 1.09) for the lateral compartment and 3.44 (SD = 0.74) for the medial and 3.96 (SD = 0.96) for the lateral compartment in group B. The only parameter to show a statistically significant difference between groups was RSS(medial) (p = 0.04). Conclusion: A difference in medial subchondral bone sclerosis between two age groups was demonstrated in the absence of MR features of OA. This may represent the earliest OA change detectable on MR imaging.
    Article · Jul 2014
    • Moreover, assessment of other structures in the knee with PD–FSE fat sat provides a better spatial resolution than the more heavily T2-weighted FSE fat sat and inversion recovery sequences [41]. Gradient-recalled echo (GRE)-type sequences, even with robust fat suppression or water excitation, are insensitive to diffuse marrow abnormalities because of trabecular magnetic susceptibility and will not show the full extent of this lesion424344. Yao and Lee were the first to describe a series of eight patients with acute knee injury and with normal radiographs but in whom MR showed irregular foci of increased signal on T2-W and low signal on T1-W spin-echo (SE) images [45].
    [Show abstract] [Hide abstract] ABSTRACT: Bone bruises are focal abnormalities in subchondral bone marrow due to trabecular microfractures as a result of traumatic force. These trauma-induced lesions are better detected with magnetic resonance (MR) imaging using water-sensitive sequences. Moreover, the pattern of bone bruise is distinctive and allows us to understand the dynamics of trauma and to predict associated soft injuries. This article discusses the mechanism of traumatic injury and MR findings.
    Full-text · Article · Aug 2013
    • Additionally, ligamentous laxity/rupture and interposition of synovial tissue or joint effusion between articular surfaces can decrease the accuracy of XR JSN as a measure of cartilage loss (Figure 5). Thus, the ability of MRI to visualize articular cartilage directly rather than only on the basis of the width of the space between opposing articular cortices is a substantial advantage2122232425. The MRI protocol used in this study is the same as that used for monitoring bone erosion with RAMRIS in many other clinical trials of RA12345.
    [Show abstract] [Hide abstract] ABSTRACT: Introduction Magnetic resonance imaging (MRI) is increasingly being used in clinical trials of rheumatoid arthritis (RA) because of its superiority over x-ray radiography (XR) in detecting and monitoring change in bone erosion, osteitis and synovitis. However, in contrast to XR, the MRI scoring method that was used in most clinical trials did not include cartilage loss. This limitation has been an obstacle to accepting MRI as a potential alternative to XR in clinical trials. Cross-sectional studies have shown MRI to be sensitive for cartilage loss in the hands and wrist; although, longitudinal sensitivity to change has not yet been confirmed. In this study we examined the ability of MRI to monitor change in cartilage loss in patients with RA in a multi-site clinical trial setting. Methods Thirty-one active RA patients from a clinical trial (IMPRESS) who were randomized equally into treatment with either rituximab + methotrexate or placebo + methotrexate had MRI of the dominant hand/wrist at baseline, 12 weeks and 24 weeks at 3 clinical sites in the US. Twenty-seven of these patients also had XR of both hands/wrists and both feet at baseline and 24 weeks. One radiologist scored all XR images using the van der Heijde-modified Sharp method blinded to visit order. The same radiologist scored MR images for cartilage loss using a previously validated 9-point scale, and bone erosion using the Outcome Measures in Rheumatology Clinical Trials (OMERACT) RA MRI Score (RAMRIS) blinded to visit order and XR scores. Data from the two treatment arms were pooled for this analysis. Results Mean MRI cartilage score increased at 12 and 24 weeks, and reached statistical significance at 24 weeks. XR total Sharp score, XR erosion score and XR joint-space narrowing (JSN) score all increased at 24 weeks, but only XR total Sharp score increased significantly. Conclusions To our knowledge, this is the first publication of a study demonstrating MRI's ability to monitor cartilage loss in a multi-site clinical trial. Combined with MRI's established performance in monitoring bone erosions in RA, these findings suggest that MRI may offer a superior alternative to XR in multi-site clinical trials of RA.
    Full-text · Article · Mar 2013
Show more