Download Knee Joint: Comprehensive Assessment with 3D Isotropic

䡲 MUSCULOSKELETAL IMAGING
Note: This copy is for your personal non-commercial use only. To order presentation-ready
copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.
ORIGINAL RESEARCH
Knee Joint: Comprehensive
Assessment with 3D Isotropic
Resolution Fast Spin-Echo MR
Imaging—Diagnostic
Performance Compared with That
of Conventional MR Imaging at
3.0 T1
Richard Kijowski, MD
Kirkland W. Davis, MD
Michael A. Woods, MD
Mary J. Lindstrom, PhD
Arthur A. De Smet, MD
Garry E. Gold, MD
Reed F. Busse, PhD
Purpose:
Materials and
Methods:
Results:
1
From the Departments of Radiology (R.K., K.W.D.,
M.A.W., A.A.D.S.) and Biostatistics (M.J.L.), University of
Wisconsin, Clinical Science Center-E3/311, 600 Highland
Ave, Madison, WI 53792; Departments of Radiology, Bioengineering, and Orthopedic Surgery, Stanford University,
Stanford, Calif (G.E.G.); and Global Applied Science Laboratory, GE Healthcare, Madison, Wis (R.F.B.). From the
2008 RSNA Annual Meeting. Received January 6, 2009;
revision requested February 18; revision received March
16; final version accepted March 30. Address correspondence to R.K. (e-mail: [email protected] ).
姝 RSNA, 2009
486
Conclusion:
To determine whether a three-dimensional isotropic resolution fast spin-echo sequence (FSE-Cube) has similar diagnostic performance as a routine magnetic resonance
(MR) imaging protocol for evaluating the cartilage, ligaments, menisci, and osseous structures of the knee joint in
symptomatic patients at 3.0 T.
This prospective, HIPAA-compliant, institutional review board–
approved study was performed with a waiver of informed consent. FSE-Cube was added to the routine 3.0-T MR imaging
protocol performed in 100 symptomatic patients (54 male patients with a median age of 32 years and 46 female patients with
a median age of 33 years) who subsequently underwent arthroscopic knee surgery. All MR imaging studies were independently reviewed twice by two musculoskeletal radiologists. During the first review, the routine MR imaging protocol was used
to detect cartilage lesions, ligament tears, meniscal tears, and
bone marrow edema lesions. During the second review, FSECube with multiplanar reformations was used to detect these
joint abnormalities. With arthroscopic results as the reference
standard, the sensitivity and specificity of FSE-Cube and the
routine MR imaging protocol in the detection of cartilage lesions, anterior cruciate ligament tears, and meniscal tears were
calculated. Permutation tests were used to compare sensitivity
and specificity values.
FSE-Cube had significantly higher sensitivity (P ⫽ .039)
but significantly lower specificity (P ⫽ .003) than the routine MR imaging protocol for detecting cartilage lesions.
There were no significant differences (P ⫽ .183–.999) in
sensitivity and specificity between FSE-Cube and the routine MR imaging protocol in the detection of anterior cruciate ligament tears, medial meniscal tears, or lateral meniscal tears. FSE-Cube depicted 96.2% of medial collateral
ligament tears, 100% of lateral collateral ligament tears,
and 85.3% of bone marrow edema lesions identified on
images obtained with the routine MR imaging protocol.
FSE-Cube has similar diagnostic performance as a routine
MR imaging protocol for detecting cartilage lesions, cruciate
ligament tears, collateral ligament tears, meniscal tears, and
bone marrow edema lesions within the knee joint at 3.0 T.
娀 RSNA, 2009
radiology.rsnajnls.org ▪ Radiology: Volume 252: Number 2—August 2009
MUSCULOSKELETAL IMAGING: Assessment of Knee Joint with FSE-Cube
T
hree-dimensional (3D) sequences
with isotropic resolution have the
potential to improve the quality
and efficiency of musculoskeletal magnetic resonance (MR) imaging. Current
musculoskeletal MR imaging protocols
can be time consuming and often consist
of two-dimensional (2D) fast spin-echo
(FSE) sequences repeated in multiple
planes. Although these 2D sequences
have high in-plane spatial resolution, they
have relatively thick sections and gaps between sections that can lead to partialvolume artifacts. Three-dimensional isotropic resolution sequences can reduce
partial-volume artifacts through the acquisition of thin continuous sections
through joints. Furthermore, the isotropic source data can be used to create multiplanar reformations (MPRs), thereby
eliminating the need to repeat sequences
with identical tissue contrast in multiple
planes. The use of 3D isotropic resolution
sequences in clinical practice could markedly decrease MR imaging examination
times, which would improve patient comfort, reduce motion artifacts, and increase the clinical efficiency of the MR
imaging unit.
Until recently, the use of 3D sequences with isotropic resolution in musculoskeletal MR imaging has been limited
by their long acquisition and postprocessing times (1,2). However, with the development of more efficient imaging techniques and the availability of high-performance MR imaging workstations,
Advances in Knowledge
䡲 A 5-minute sagittal fast spin-echo
(FSE)-Cube sequence with multiplanar reformations (MPRs) has
higher sensitivity but lower specificity than a routine MR imaging
protocol in the detection of cartilage lesions within the knee joint
at 3.0 T.
䡲 A 5-minute sagittal FSE-Cube sequence with MPRs has similar
sensitivity and specificity as a routine MR imaging protocol for detecting ligament tears, meniscal
tears, and bone marrow edema
lesions within the knee joint at
3.0 T.
evaluating the knee joint by using 3D isotropic resolution sequences has become
clinically feasible (3–5). Most currently
used 3D sequences with isotropic resolution are balanced steady-state free precession sequences (1–5). Three-dimensional FSE sequences with intermediateweighted contrast have also been
developed, but their use in clinical practice is currently limited by their anisotropic resolution and relatively long acquisition times (6).
FSE-Cube is a new 3D FSE sequence
that can be used to evaluate the knee
joint. FSE-Cube can produce multiplanar
3D intermediate-weighted images with
0.7-mm isotropic resolution at 3.0 T after
a single 5-minute acquisition (7,8). The
use of FSE-Cube for multiplanar evaluation of the knee joint in asymptomatic
volunteers has been previously reported
(9). However, clinical studies with arthroscopic correlation are needed to assess
the strengths and weaknesses of FSECube and to determine whether the 3D
sequence can replace currently used 2D
sequences for evaluating the knee joint in
clinical practice. Thus, this study was performed to determine whether FSE-Cube
has similar diagnostic performance as a
routine MR imaging protocol for evaluating the cartilage, ligaments, menisci, and
osseous structures of the knee joint in
symptomatic patients at 3.0 T.
Materials and Methods
Study Group
One author (R.F.B.) is an employee of GE
Healthcare (Waukesha, Wisconsin), and
another author (G.E.G.) receives research support from GE Healthcare.
These authors did not have control of inclusion of any data or information. The
study was performed in compliance with
Health Insurance Portability and Accountability Act regulations, with apImplication for Patient Care
䡲 A sagittal FSE-Cube sequence
with MPRs has the potential to
provide rapid comprehensive assessment of the knee joint in
symptomatic patients at 3.0 T.
Radiology: Volume 252: Number 2—August 2009 ▪ radiology.rsnajnls.org
Kijowski et al
proval from the institutional review board
of the University of Wisconsin, and with a
waiver of informed consent.
Between December 1, 2007, and October 1, 2008, 272 symptomatic patients
(143 male patients [age range, 16 –72
years; median age, 35 years] and 129 female patients [age range, 16 – 81 years;
median age, 36 years]) undergoing routine MR imaging of the knee at our institution were enrolled in a prospective clinical study investigating the ability of FSECube to provide comprehensive knee
joint assessment. All 272 patients enrolled in the study were imaged with our
routine knee MR imaging protocol, which
consists of multiplanar 2D FSE sequences, and the FSE-Cube sequence.
One hundred of these 272 patients subsequently underwent arthroscopic knee
surgery. The study group consisted of
these 100 consecutive patients (54 male
patients [age range, 16 – 66 years; median
age, 32 years] and 46 female patients [age
range, 16 –79 years; median age, 33
years]) who were evaluated with MR imaging and arthroscopy. No patient was
excluded from the study on the basis of
any factor, including age, weight, severity
of knee injury, history of prior knee surgery, or quality of the MR imaging examination.
MR Imaging
All 100 patients in the study group were
imaged with the same 3.0-T MR imaging
Published online
10.1148/radiol.2523090028
Radiology 2009; 252:486 – 495
Abbreviations:
FSE ⫽ fast spin echo
MPR ⫽ multiplanar reformation
3D ⫽ three-dimensional
2D ⫽ two-dimensional
Author contributions:
Guarantor of integrity of entire study, R.K.; study concepts/study design or data acquisition or data analysis/
interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors;
literature research, R.K., M.A.W., G.E.G.; clinical studies,
R.K., K.W.D., M.A.W., A.A.D.S.; statistical analysis, R.K.,
M.A.W., M.J.L.; and manuscript editing, R.K., A.A.D.S.,
G.E.G.
See Materials and Methods for pertinent disclosures.
487
MUSCULOSKELETAL IMAGING: Assessment of Knee Joint with FSE-Cube
unit (Sigma Excite HDx; GE Healthcare)
by using an eight-channel phased-array
extremity coil (Precision Eight TX/TR
High Resolution Knee Array; Invivo, Orlando, Fla). All MR imaging examinations
consisted of an axial frequency selective
fat-suppressed T2-weighted FSE sequence, a coronal intermediate-weighted
FSE sequence, a coronal frequency selective fat-suppressed intermediateweighted FSE sequence, a sagittal intermediate-weighted FSE sequence, a sagittal frequency selective fat-suppressed
T2-weighted FSE sequence, and a sagittal
FSE-Cube sequence. The imaging parameters of all sequences are summarized in
Table 1. The FSE-Cube sequence was
performed with a 2D autocalibrating parallel imaging reconstruction technique
(ARC; GE Healthcare) with an acceleration factor of three to reduce imaging
time. The FSE-Cube isotropic source data
were used to create sagittal, coronal, and
axial reformatted images of the knee joint
with 1.5-mm section thickness. The postprocessing was performed by a technologist on the imaging workstation immediately after the MR imaging examination.
Arthroscopic Knee Surgery
Arthroscopic knee surgery was performed in all 100 patients in the study
group within 3 months (time range, 3– 81
days; mean, 37.3 days ⫾ 18.0 [standard
deviation]) of their MR imaging examination. All arthroscopic knee surgeries were
performed by one of three experienced
orthopedic surgeons at the University of
Wisconsin who specialized in sports med-
Kijowski et al
icine and who had between 10 and 25
years of clinical experience. The decision
to perform arthroscopic surgery was
based on clinical findings and the official
interpretations of the MR imaging studies. The official interpretations of the MR
imaging studies were made by one of
seven fellowship-trained musculoskeletal
radiologists at our institution (including
R.K., K.W.D., and A.A.D.S.) by using
the routine MR imaging protocol, as mandated by our internal review board.
All articular surfaces of the knee joint
were graded at arthroscopy by using the
Noyes classification system (grade 0 ⫽
normal, grade 1 ⫽ cartilage softening,
grade 2A ⫽ superficial partial-thickness
cartilage lesion ⬍ 50% of the total thickness of the articular surface, grade 2B ⫽
deep partial-thickness cartilage lesion ⬎
50% of the total thickness of the articular
surface, and grade 3 ⫽ full-thickness cartilage lesion) (10). The presence of anterior cruciate ligament tears, posterior
cruciate ligament tears, and meniscal
tears was also documented at arthroscopy. The anterior cruciate ligament
tears and posterior cruciate ligament
tears were classified as partial thickness
or full thickness. No attempt was made to
classify the meniscal tears. The orthopedic surgeons were aware of the official
interpretations of the MR imaging studies
in all patients at the time of arthroscopy.
Review of MR Imaging Studies
All MR imaging studies were independently
reviewed twice in separate sittings by two
fellowship-trained musculoskeletal radiolo-
gists (R.K. and K.W.D., with 10 and 14
years of clinical experience, respectively).
The radiologists were unaware of the arthroscopic findings in each patient when
they reviewed the MR imaging studies. To
prevent recall bias, the radiologists reviewed the MR imaging studies in separate
sittings at least 4 months apart.
During the first review of the MR imaging studies, the radiologists used all the
sequences in the routine MR imaging protocol together to detect cartilage lesions,
anterior cruciate ligament tears, posterior cruciate ligament tears, meniscal
tears, medial collateral ligament tears,
lateral collateral ligament tears, and bone
marrow edema lesions within the knee
joint. During the second review, the radiologists used the FSE-Cube sequence with
MPRs to detect these joint abnormalities.
During both reviews, the radiologists
graded all articular surfaces of the knee
joint by using a modified Noyes classification system (grade 0 ⫽ normal cartilage,
grade 1 ⫽ increased T2 signal intensity of
morphologically normal cartilage, grade
2A ⫽ superficial partial-thickness cartilage lesion ⬍ 50% of the total thickness of
the articular surface, grade 2B ⫽ deep
partial-thickness cartilage lesion ⬎ 50%
of the total thickness of the articular surface, and grade 3 ⫽ full-thickness cartilage lesion) (11–13). No attempt was
made to classify the anterior cruciate ligament tears, posterior cruciate ligament
tears, or meniscal tears.
A third review of the MR imaging
studies was performed by both radiologists to obtain a consensus interpretation
Table 1
Parameters for MR Imaging Sequences
Parameter
Repetition time (msec)
Echo time (msec)
Matrix size
Field of view (cm)
Section thickness (mm)
Bandwidth (kHz)
Echo train length
No. of signals acquired
Imaging time
488
Axial Fat-suppressed
T2-weighted FSE
Sequence
Coronal
Intermediate-weighted
FSE Sequence
Coronal Fat-suppressed
Intermediate-weighted
FSE Sequence
Sagittal
Intermediate-weighted
FSE Sequence
Sagittal Fat-suppressed
T2-weighted FSE
Sequence
Sagittal
FSE-Cube
Sequence
4300
77
448 ⫻ 224
18
3
41.7
21
4
3 Min 30 sec
1800
20
384 ⫻ 224
14
2
31.2
4
2
3 Min 25 sec
2000
20
384 ⫻ 224
14
2
31.2
4
2
3 Min 26 sec
2000
20
384 ⫻ 224
14
2
31.2
4
2
3 Min 26 sec
5300
80
384 ⫻ 224
14
3
41.7
20
3
3 Min 16 sec
2200
24
224 ⫻ 224
15
0.7
31.2
44
0.5
5 Min
radiology.rsnajnls.org ▪ Radiology: Volume 252: Number 2—August 2009
MUSCULOSKELETAL IMAGING: Assessment of Knee Joint with FSE-Cube
of the images acquired with the routine
MR imaging protocol in terms of the presence or absence of medial collateral ligament tears, lateral collateral ligaments
tears, and bone marrow edema lesions.
This consensus interpretation was used
as the reference standard to determine
the diagnostic performance of FSE-Cube
for detecting these joint abnormalities.
The medical records of all patients with a
consensus interpretation of medial collateral ligament tears or lateral collateral ligament tears with the routine MR imaging
protocol were reviewed. All 15 patients
with MR imaging findings of a medial collateral ligament tear had medial joint line
pain and tenderness at palpation of the
medial collateral ligament. Two patients
also had valgus instability at examination
with anesthesia. All five patients with MR
imaging findings of a lateral collateral ligament tear had lateral joint line pain and
tenderness at palpation of the lateral collateral ligament.
Statistical Analysis
All statistical analyses were performed
by using the R programming environment (R: A Language and Environment
for Statistical Computing, version 2.3.1;
R Foundation of Statistical Imaging, Vienna, Austria, 2006 [http://www.r-project
.org]). For all statistical tests, differences were considered to be significant if the P value was less than .05.
Statistical analysis was used to compare the demographic data of the 100
patients in the study group who underwent arthroscopic knee surgery with
those of the 187 patients enrolled in the
prospective clinical study who did not
undergo arthroscopic knee surgery and
who were thus excluded from the study
group. The Wilcoxon test was used to
compare the median ages of the male
and female patients. ␹2 Tests were used
to compare the proportions of male and
female patients and the proportions of
patients with right and those with left
knee pain.
Statistical analysis was used to compare the demographic characteristics of
the 54 male patients and 46 female patients in the study group. The Wilcoxon
test was used to compare the median age.
We used t tests to compare the mean
time interval between MR imaging and
arthroscopic knee surgery. ␹2 Tests were
used to compare the proportions of patients with right and those with left knee
pain, the proportions of patients undergoing arthroscopic surgery performed by
each orthopedic surgeon, and the proportions of patients with each surgical indication (ie, anterior cruciate ligament reconstruction, medial meniscus repair or
resection, lateral meniscus repair or resection, and osteochondral autograft
transplantation).
At statistical analysis to assess the diagnostic performance of FSE-Cube and
the routine MR imaging protocol for detecting cartilage lesions, ligament tears,
meniscal tears, and bone marrow edema
lesions within the knee joint, the data
from the independent reviews of both
readers were combined. This was done to
increase statistical power for a comparison between FSE-Cube and the routine
MR imaging protocol.
With arthroscopy as the reference
standard, the sensitivity, specificity, and
accuracy of FSE-Cube and the routine MR
imaging protocol for detecting each grade
and all grades of cartilage lesions within
the knee joint were calculated for both
readers combined and for all articular
surfaces combined. For calculating sensitivity, specificity, and accuracy, the cartilage grades assigned at MR imaging were
classified as either “disease negative” (ie,
MR imaging grade 0) or “disease positive”
(ie, MR imaging grades 1, 2A, 2B, and 3).
The proportions of cartilage lesions
graded identically and the proportions of
cartilage lesions graded within one grade
at arthroscopy and at MR imaging were
calculated for FSE-Cube and the routine
MR imaging protocol for both readers
combined. Standard errors of the mean
were calculated by “bootstrapping” patients to account for dependence within
patients among the six articular surfaces
and among the two readers. On the basis
of the standard errors of the mean, permutation tests were used to compare differences between FSE-Cube and the routine MR imaging protocol.
With arthroscopy as the reference
standard, the sensitivity, specificity, and
accuracy of FSE-Cube and the routine MR
imaging protocol for detecting anterior
Radiology: Volume 252: Number 2—August 2009 ▪ radiology.rsnajnls.org
Kijowski et al
cruciate ligament tears, posterior cruciate ligament tears, medial meniscal tears,
and lateral meniscal tears within the knee
joint were calculated for both readers
combined. On the basis of standard errors of the mean calculated by “bootstrapping” patients, permutation tests were
used to compare differences between
FSE-Cube and the routine MR imaging
protocol. Using the consensus interpretations of the routine MR imaging data as
the reference standard, the sensitivity,
specificity, and accuracy of FSE-Cube for
detecting medial collateral ligament tears,
lateral collateral ligament tears, and bone
marrow edema lesions were calculated
for both readers combined.
␬ Statistics were used to measure interobserver agreement between readers
for determining the presence or absence
of cartilage lesions, anterior cruciate ligament tears, posterior cruciate ligament
tears, medial meniscal tears, lateral meniscal tears, medial collateral ligament
tears, lateral collateral ligament tears,
and bone marrow edema lesions. Interobserver agreement was assessed according to the recommendations of Landis and Koch (14), in which a ␬ value of
0.00 – 0.20 indicates slight agreement; a ␬
value of 0.21– 0.40, fair agreement; a ␬
value of 0.41– 0.60, moderate agreement;
a ␬ value of 0.61– 0.80, substantial agreement; a ␬ value of 0.81 to less than 1.00,
almost perfect agreement; and a ␬ value
of 1.00, perfect agreement. On the basis
of standard errors of the mean calculated
by “bootstrapping” patients, permutation
tests were used to compare differences
between FSE-Cube and the routine MR
imaging protocol.
Results
There were no significant differences between the 100 patients in the study group
who underwent arthroscopic knee surgery and the 187 patients enrolled in the
prospective clinical study who did not undergo arthroscopic knee surgery with regard to the median age of male patients
(P ⫽ .09), the median age of female patients (P ⫽ .23), the proportions of male
and female patients (P ⫽ .82), and the
proportions of patients with right and
those with left knee pain (P ⫽ .96).
489
MUSCULOSKELETAL IMAGING: Assessment of Knee Joint with FSE-Cube
There were no significant differences
between the 54 male patients and the 46
female patients in the study group with
regard to median age (P ⫽ .89), the mean
time interval between MR imaging and
arthroscopic knee surgery (P ⫽ .47), the
proportions of patients with right and
those with left knee pain (P ⫽ .89), the
proportions of patients undergoing arthroscopic surgery performed by each orthopedic surgeon (P ⫽ .96), and the proportions of patients with each surgical indication (P ⫽ .91–.99).
As shown in Table 2, the sensitivity,
Kijowski et al
specificity, and accuracy, respectively,
for detecting 189 cartilage lesions within
the knee joint were 72.8%, 88.2%, and
83.3% for FSE-Cube and 68.2%, 92.8%,
and 85.1% for the routine MR imaging
protocol (Figs 1 and 2). FSE-Cube had
significantly higher sensitivity (P ⫽
.039), significantly lower specificity
(P ⫽ .003), and similar accuracy (P ⫽
.062) compared with the routine MR
imaging protocol for detecting cartilage lesions. There was no significant
difference between FSE-Cube and the
routine MR imaging protocol in the
proportions of cartilage lesions graded
identically (33.9% [128 of 378] for
FSE-Cube and 28.8% [109 of 378] for
the routine MR imaging protocol, P ⫽
.228) or within one grade of the arthroscopic grade (72.0% [272 of 378]
for FSE-Cube and 68.0% [257 of 378]
for the routine MR imaging protocol,
P ⫽ .203).
As shown in Table 3, there were no
statistically significant differences (P ⫽
.183–.999) in sensitivity, specificity, and
accuracy between FSE-Cube and the routine MR imaging protocol for detecting 33
Table 2
Sensitivity, Specificity, and Accuracy of FSE-Cube and Routine MR Imaging Protocol in Detection of Cartilage Lesions within Knee
Joint for Both Readers Combined and for All Articular Surfaces Combined
Cartilage Lesions
Grade 1 (n ⫽ 10)
Grade 2A (n ⫽ 56)
Grade 2B (n ⫽ 88)
Grade 3 (n ⫽ 35)
All lesions combined
(n ⫽ 189)
FSE-Cube
Sensitivity (%)
Routine MR Imaging
Protocol
FSE-Cube
Specificity (%)
Routine MR Imaging
Protocol
FSE-Cube
Accuracy (%)
Routine MR Imaging
Protocol
25.0 (5/20) [.200]
51.0 (57/112) [.116]
84.6 (149/176) [.040]
91.4 (64/70) [.124]
25.0 (5/20)
46.3 (52/112)
78.9 (139/176)
88.6 (62/70)
88.2 (725/822) [.003]
88.2 (725/822) [.003]
88.2 (725/822) [.003]
88.2 (725/822) [.003]
92.8 (763/822)
92.8 (763/822)
92.8 (763/822)
92.8 (763/822)
86.7 (730/842) [.006]
83.7 (782/934) [.020]
87.6 (874/998) [.018]
88.4 (789/892) [.008]
91.2 (768/842)
87.2 (815/934)
90.4 (902/998)
92.4 (825/892)
72.8 (275/378) [.039]
68.2 (258/378)
88.2 (725/822) [.003]
92.8 (763/822)
83.3 (1000/1200) [.062]
85.1 (1021/1200)
Note.—Data in parentheses were used to calculate the percentages (numerators and denominators represent the combined data from the independent reviews of the two readers); data in brackets
are P values for comparison of the two imaging techniques. P ⬍ .05 indicates a significant difference.
Figure 1
Figure 1: Sagittal MR images in 43-year-old man with surgically confirmed grade 2B cartilage lesion on the femoral trochlea that was detected by neither reader with
the routine MR imaging protocol and by both readers with FSE-Cube. (a) Intermediate-weighted FSE image and (b) fat-suppressed T2-weighted FSE image of knee joint
show normal-appearing articular cartilage on the femoral trochlea (arrow). (c) Corresponding FSE-Cube image shows a deep partial-thickness cartilage lesion on the
femoral trochlea (arrow).
490
radiology.rsnajnls.org ▪ Radiology: Volume 252: Number 2—August 2009
MUSCULOSKELETAL IMAGING: Assessment of Knee Joint with FSE-Cube
full-thickness and three partial-thickness
anterior cruciate ligament tears, 52 medial meniscal tears, and 35 lateral meniscal tears (Figs 3–5). FSE-Cube also had
similar specificity (P ⫽ .999) and accuracy (P ⫽ .999) compared with the routine MR imaging protocol for detecting
posterior cruciate ligament tears. When
the consensus interpretation of the routine MR imaging data was used as the
reference standard, the sensitivities of
FSE-Cube for detecting 13 medial collateral ligament tears, five lateral collateral
ligament tears, and 149 bone marrow
edema lesions were 96.2%, 100%, and
85.3%, respectively (Table 3, Fig 6). FSE-
Kijowski et al
Cube had high specificity and accuracy in
the detection of these joint abnormalities.
As shown in Table 4, there was no
significant difference (P ⫽ .073–.999) in
interobserver agreement between FSECube and the routine MR imaging protocol for determining the presence or absence of cartilage lesions, anterior cruciate ligament tears, posterior cruciate
ligament tears, medial meniscal tears, lateral meniscal tears, medial collateral ligament tears, lateral collateral ligament
tears, and bone marrow edema lesions
within the knee joint. Both FSE-Cube and
the routine MR imaging protocol had interobserver agreement that ranged be-
tween moderate and perfect for determining the presence or absence of these
joint abnormalities.
Discussion
Three-dimensional sequences with isotropic resolution are commonly balanced
steady-state free precession sequences
(1–5). Results of two recent studies (3,5)
have shown that these sequences can be
used to provide rapid comprehensive
knee joint assessment in symptomatic patients. In a preliminary study performed
in 30 patients with arthroscopic correlation, Duc and colleagues (3) found that
Figure 2
Figure 2: Sagittal MR images in 41-year-old woman with surgically confirmed normal articular cartilage on the lateral femoral condyle that was detected by both readers with the routine MR imaging protocol and by neither reader with FSE-Cube. (a) Intermediate-weighted FSE image and (b) fat-suppressed T2-weighted FSE image of
knee joint show a smooth articular surface of the lateral femoral condyle (arrow). (c) Corresponding FSE-Cube image shows an irregular articular surface of the lateral
femoral condyle (arrow); this was thought by both readers to represent a superficial partial-thickness cartilage lesion.
Table 3
Sensitivity, Specificity, and Accuracy of FSE-Cube and Routine MR Imaging Protocol in Detection of Knee Joint Abnormalities for Both
Readers Combined
Joint Abnormality*
ACL tear (n ⫽ 36)
PCL tear (n ⫽ 0)
MM tear (n ⫽ 52)
LM tear (n ⫽ 35)
MCL tear (n ⫽ 13)
LCL tear (n ⫽ 5)
BME lesion (n ⫽ 149)
FSE-Cube
Sensitivity (%)
Routine MR Imaging
Protocol
100 (72/72) [.999]
NA
98.1 (102/104) [.615]
72.9 (51/70) [.353]
96.2 (25/26) [NA]
100 (10/10) [NA]
85.3 (250/193) [NA]
100 (72/72)
NA
97.1 (101/104)
80.0 (56/70)
NA
NA
NA
FSE-Cube
Specificity (%)
Routine MR Imaging
Protocol
98.4 (126/128) [.999]
99.0 (198/200) [.999]
70.8 (68/96) [.413]
85.8 (111/130) [.183]
96.6 (168/174) [NA]
97.3 (184/190) [NA]
95.0 (862/907) [NA]
98.4 (126/128)
99.5 (199/200)
65.6 (63/96)
79.2 (103/130)
NA
NA
NA
FSE-Cube
Accuracy (%)
Routine MR Imaging
Protocol
99.0 (198/200) [.999]
99.0 (198/200) [.999]
85.0 (170/200) [.450]
81.0 (162/200) [.694]
96.6 (193/200) [NA]
97.5 (194/200) [NA]
92.7 (1112/1200) [NA]
99.0 (198/200)
99.5 (199/200)
82.0 (164/200)
79.5 (159/200)
NA
NA
NA
Note.—Data in parentheses were used to calculate the percentages (numerators and denominators represent the combined data from the independent reviews of the two readers); data in brackets
are P values for comparison of the two imaging techniques. P ⬍ .05 indicates a significant difference. NA ⫽ not applicable.
* ACL ⫽ anterior cruciate ligament, BME ⫽ bone marrow edema, LCL ⫽ lateral collateral ligament, LM ⫽ lateral meniscus, MCL ⫽ medial collateral ligament, MM ⫽ medial meniscus, PCL ⫽
posterior cruciate ligament.
Radiology: Volume 252: Number 2—August 2009 ▪ radiology.rsnajnls.org
491
MUSCULOSKELETAL IMAGING: Assessment of Knee Joint with FSE-Cube
isotropic resolution water excitation true
fast imaging with steady-state precession
had similar sensitivity and specificity as a
routine MR imaging protocol for detecting cartilage lesions, anterior cruciate ligament tears, and meniscal tears. However, a larger study performed by Kijowski and associates (5) described
potential limitations of the T2/T1weighted tissue contrast of balanced
steady-state free precession sequences.
In that study, isotropic resolution vastly
undersampled isotropic projection steadystate free precession was found to have
similar sensitivity and specificity as a
routine MR imaging protocol for detecting cartilage lesions, ligament tears, and
medial meniscal tears in 95 patients
with arthroscopic correlation but to
have significantly lower (P ⬍ .05) sensitivity for detecting lateral meniscal tears
and bone marrow edema lesions.
Three-dimensional FSE sequences
with isotropic resolution provide another
promising method to replace currently
used 2D sequences in clinical practice.
Three-dimensional FSE sequences can
produce images with intermediateweighted contrast, which is the most
commonly used tissue contrast in musculoskeletal MR imaging (13,15–22). In our
study, FSE-Cube with MPRs provided
similar clinical information regarding the
cartilage, ligaments, menisci, and osseous
structures of the knee joint as an entire
25-minute routine MR imaging protocol.
To our knowledge, only one previous
group (6) has reported the diagnostic performance of a 3D FSE sequence for evaluating the knee joint in symptomatic patients. However, that study was limited
by a relatively small patient population
and the use of a 3D intermediateweighted sequence with anisotropic resolution (0.6 ⫻ 0.6 ⫻ 1.0-mm voxel size).
In our study, FSE-Cube had significantly higher sensitivity but significantly
lower specificity than the routine MR
imaging protocol for detecting cartilage
lesions within the knee joint. To our
knowledge, no previous group has reported the diagnostic performance of a
3D FSE sequence for evaluating the
knee articular cartilage. The higher sensitivity of FSE-Cube for detecting cartilage lesions was most likely secondary
492
Kijowski et al
Figure 3
Figure 4
Figure 3: Sagittal MR images in 19-year-old
man with surgically confirmed anterior cruciate
ligament tear that was detected by both readers
with the routine MR imaging protocol and FSECube. (a) Fat-suppressed T2-weighted FSE image
and (b) corresponding FSE-Cube image of knee
joint show complete disruption of the fibers of the
anterior cruciate ligament (arrow).
to reduced partial-volume averaging.
The FSE-Cube images had thinner section thicknesses than the 2D FSE images, with no gaps between sections;
this likely provided better visibility of
small cartilage lesions. The lower specificity of FSE-Cube was most likely secondary to decreased in-plane spatial
resolution and image blurring due to acquisition of high spatial frequencies late
in the echo train. These factors may
cause a normal articular surface to appear indistinct and ill defined, simulating the appearance of superficial cartilage degeneration.
FSE-Cube had similar sensitivity,
specificity, and accuracy as the routine
Figure 4: Sagittal MR images in 25-year-old
man with surgically confirmed tear of the posterior
horn of the medial meniscus that was detected by
both readers with the routine MR imaging protocol
and FSE-Cube. (a) Intermediate-weighted FSE
image and (b) corresponding FSE-Cube image of
knee joint show a tear of the posterior horn of the
medial meniscus (arrow) with an associated parameniscal cyst (arrowhead).
MR imaging protocol for evaluating the
ligaments of the knee joint. Yoon and colleagues (6) also found that a 3D intermediate-weighted FSE sequence with anisotropic resolution had high diagnostic performance for detecting 10 anterior
cruciate ligament tears and one posterior
cruciate ligament tear confirmed at arthroscopy. FSE-Cube may have advantages over 2D sequences for evaluating
the knee ligaments. The thin, continuous
sections of FSE-Cube minimize the effect
of partial-volume averaging, which can be
a source of diagnostic error when evaluating the anterior cruciate ligament (17).
In addition, the isotropic resolution of
FSE-Cube allows reformatted images to
be created in any orientation after a single
acquisition. As a result, coronal oblique
radiology.rsnajnls.org ▪ Radiology: Volume 252: Number 2—August 2009
MUSCULOSKELETAL IMAGING: Assessment of Knee Joint with FSE-Cube
and sagittal oblique images, which are especially useful for evaluating the ligaments of the posteromedial and posterolateral corners of the knee (23–25), can
be obtained.
FSE-Cube had similar sensitivity,
specificity, and accuracy as the routine
MR imaging protocol for evaluating the
menisci of the knee joint. Yoon and colleagues (6) also found that a 3D intermediate-weighted FSE sequence with
anisotropic resolution had high diagnostic performance for detecting 26 meniscal tears confirmed at arthroscopy. A
potential disadvantage of FSE sequences such as FSE-Cube for evaluating the menisci is image blurring
(26,27). Despite the use of flip angle
modulation to constrain T2 decay over
an extended echo train, FSE-Cube im-
ages suffer from blurring at subjective
analysis because of the acquisition of
high spatial frequencies late in the echo
train (28). Blurring on FSE-Cube images can be reduced by decreasing echo
train length or increasing bandwidth at
the expense of acquisition time and signal-to-noise ratio. The interaction between image blurring, acquisition time,
bandwidth, and signal-to-noise ratio is
complex, with many combinations possible. Additional studies are needed to
determine how the bandwidth, echo
train length, and parallel imaging acceleration of FSE-Cube should be optimized to minimize blurring while main-
Figure 6
Figure 5
Kijowski et al
taining adequate signal-to-noise ratio
and clinically feasible imaging times.
FSE-Cube had similar sensitivity,
specificity, and accuracy as the routine
MR imaging protocol for detecting bone
marrow edema lesions within the knee
joint. Intermediate-weighted FSE sequences such as FSE-Cube may require
fat suppression to provide optimal visualization of bone marrow edema lesions
(15). Fat suppression also improves detection of superficial cartilage lesions by
increasing the contrast between articular
cartilage and synovial fluid and improves
visualization of edema and hemorrhage
within injured ligaments (18,21,22). FSECube uses a spectral inversion recovery
pulse for fat suppression that does not
prolong acquisition time or reduce anatomic coverage. This is because the majority of each repetition time is spent in
signal recovery rather than in acquiring
data from adjacent sections (9).
FSE-Cube and the routine MR imaging protocol had similar interobserver
agreement for determining the presence or absence of cartilage lesions, ligament tears, meniscal tears, and bone
marrow edema lesions within the knee
joint. However, the moderate interobTable 4
Interobserver Agreement with
FSE-Cube and Routine MR Imaging
Protocol for Determining Presence or
Absence of Knee Joint Abnormalities
Joint
Abnormality* FSE-Cube
Figure 5: Sagittal MR images in 23-year-old
woman show surgically confirmed tear of the posterior horn of the lateral meniscus, which was
detected by both readers with the routine MR imaging protocol and by neither reader with FSECube. (a) Intermediate-weighted FSE image of
knee joint shows a small undersurface tear of the
posterior horn of the lateral meniscus (arrow).
(b) Corresponding FSE-Cube image shows a normal-appearing undersurface of the posterior horn
of the lateral meniscus (arrow).
Figure 6: Coronal MR images in 35-year-old
man with medial collateral ligament tear that was
detected by both readers with the routine MR imaging protocol and FSE-Cube. (a) Fat-suppressed
intermediate-weighted FSE image and (b) corresponding FSE-Cube image of knee joint show
partial disruption of the proximal fibers of the medial collateral ligament (arrowhead). Also note the
subchondral bone marrow edema within the lateral
femoral condyle and lateral tibial plateau (arrows).
Radiology: Volume 252: Number 2—August 2009 ▪ radiology.rsnajnls.org
Cartilage
lesion†
ACL tear
PCL tear
MM tear
LM tear
MCL tear
LCL tear
BME lesion
0.68
0.96
1.00
0.61
0.78
0.73
0.59
0.82
Routine
MR Imaging
Protocol
P
Value
0.69
0.96
1.00
0.65
0.61
0.95
0.90
0.75
.819
.999
.999
.762
.073
.089
.155
.230
Note.—Data are ␬ values.
* ACL ⫽ anterior cruciate ligament, BME ⫽ bone marrow edema, LCL ⫽ lateral collateral ligament, LM ⫽
lateral meniscus, MCL ⫽ medial collateral ligament,
MM ⫽ medial meniscus, PCL ⫽ posterior cruciate
ligament.
†
All grades combined.
493
MUSCULOSKELETAL IMAGING: Assessment of Knee Joint with FSE-Cube
server agreement for determining the
presence or absence of medial meniscus
tears in our study was much lower than
the values observed in previous studies
(3,5,16,19). The relatively low interobserver agreement was due to the wide
difference between the two readers in
the number of false-positive interpretations at evaluation of the medial meniscus. However, the difference between
readers in the number of false-positive
interpretations was similar for both
FSE-Cube and the routine MR imaging
protocol, and this resulted in similar interobserver agreements for determining the presence or absence of medial
meniscus tears.
Our study had several limitations.
One limitation was our small patient population. Many of the P values in our study
comparing FSE-Cube and the routine MR
imaging protocol that were not statistically significant, such as those comparing
the sensitivity for detecting each grade of
cartilage lesion, the accuracy for detecting all cartilage lesions, and the sensitivity
and specificity for detecting lateral meniscal tears, were relatively low. For this
reason, we believe that additional clinical
studies with larger patient populations
may be needed to detect subtle differences in the diagnostic performance of
FSE-Cube and the routine MR imaging
protocol. In addition, multiple statistical
tests were performed in the same patient
population in our study. As a result, two
statistically significant results, such as the
higher sensitivity and lower specificity of
FSE-Cube for detecting cartilage lesions,
are not completely independent of one
another and may be partially explained by
the same underlying phenomena. Another limitation was the presence of selection bias, as our study group consisted
of only a proportion of all patients undergoing routine MR imaging of the knee at
our institution. Additional limitations included the inability to randomize the order in which the MR imaging studies were
reviewed and the inability to blind readers with regard to the MR imaging sequences they were using for comprehensive knee joint assessment. In addition,
the orthopedic surgeons were aware of
the official interpretations of the MR imaging studies when making clinical deci494
Kijowski et al
sions and performing arthroscopic surgery, which was based solely on the findings with the routine MR imaging
protocol. Our study also did not assess
the ability of FSE-Cube to help evaluate the quadriceps and patellar tendons and the ligaments of the posteromedial and posterolateral corners of the
knee. Furthermore, the consensus interpretation of the routine MR imaging data
was a less than optimal reference standard for determining the sensitivity, specificity, and accuracy of FSE-Cube in the
detection of collateral ligament tears and
bone marrow edema lesions.
In conclusion, the results of our preliminary study have shown that FSE-Cube
has similar diagnostic performance as a
routine MR imaging protocol in the detection of cartilage lesions, cruciate ligament
tears, collateral ligament tears, meniscal
tears, and bone marrow edema lesions
within the knee joint at 3.0 T. For this
reason, FSE-Cube can be used to provide
rapid comprehensive knee joint assessment in patients with severe pain or
claustrophobia who cannot tolerate a 25minute routine MR imaging examination.
However, additional studies are needed
to determine whether FSE-Cube can replace currently used 2D sequences for
evaluating the knee joint in all patients
undergoing routine MR imaging.
References
1. Gay SB, Chen NC, Burch JJ, Gleason TR,
Sagman AM. Multiplanar reconstruction in
magnetic resonance evaluation of the knee:
comparison with film magnetic resonance interpretation. Invest Radiol 1993;28(2):142–
145.
2. Wieslander SB, Rappeport ED, Lausten GS,
Thomsen HS. Multiplanar reconstruction in
MR imaging of the knee: comparison with
standard sagittal and coronal images. Acta
Radiol 1998;39(2):116 –119.
3. Duc SR, Pfirrmann CW, Koch PP, Zanetti M,
Hodler J. Internal knee derangement assessed with 3-minute three-dimensional isovoxel true FISP MR sequence: preliminary
study. Radiology 2008;246(2):526 –535.
4. Kijowski R, Lu A, Block W, Grist T. Evaluation of the articular cartilage of the knee joint
with vastly undersampled isotropic projection reconstruction steady-state free precession imaging. J Magn Reson Imaging 2006;
24(1):168 –175.
5. Kijowski R, Blankenbaker DG, Klaers JL,
Shinki K, De Smet AA, Block WF. Vastly
undersampled isotropic projection steadystate free precession imaging of the knee:
diagnostic performance compared with conventional MR. Radiology 2009;251(1):185–
194.
6. Yoon YC, Kim SS, Chung HW, Choe BK,
Ahn JH. Diagnostic efficacy in knee MRI
comparing conventional technique and multiplanar reconstruction with one-millimeter
FSE PDW images. Acta Radiol 2007;48(8):
869 – 874.
7. Busse RF, Brau AC, Vu A, et al. Effects of
refocusing flip angle modulation and view ordering in 3D fast spin echo. Magn Reson
Med 2008;60(3):640 – 649.
8. Busse RF, Hariharan H, Vu A, Brittain JH.
Fast spin echo sequences with very long echo
trains: design of variable refocusing flip angle
schedules and generation of clinical T2 contrast. Magn Reson Med 2006;55(5):1030 –
1037.
9. Gold GE, Busse RF, Beehler C, et al. Isotropic MRI of the knee with 3D fast spin-echo
extended echo-train acquisition (XETA): initial experience. AJR Am J Roentgenol 2007;
188(5):1287–1293.
10. Noyes FR, Stabler CL. A system for grading
articular cartilage lesions at arthroscopy.
Am J Sports Med 1989;17(4):505–513.
11. Bredella MA, Tirman PF, Peterfy CG, et al.
Accuracy of T2-weighted fast spin-echo MR
imaging with fat saturation in detecting cartilage defects in the knee: comparison with
arthroscopy in 130 patients. AJR Am J
Roentgenol 1999;172(4):1073–1080.
12. Potter HG, Linklater JM, Allen AA, Hannafin
JA, Haas SB. Magnetic resonance imaging of
articular cartilage in the knee: an evaluation
with use of fast-spin-echo imaging. J Bone
Joint Surg Am 1998;80(9):1276 –1284.
13. Sonin AH, Pensy RA, Mulligan ME, Hatem S.
Grading articular cartilage of the knee using
fast spin-echo proton density-weighted MR
imaging without fat suppression. AJR Am J
Roentgenol 2002;179(5):1159 –1166.
14. Landis JR, Koch GG. The measurement of
observer agreement for categorical data.
Biometrics 1977;33(1):159 –174.
15. Lal NR, Jamadar DA, Doi K, et al. Evaluation
of bone contusions with fat-saturated fast
spin-echo proton-density magnetic resonance
imaging. Can Assoc Radiol J 2000;51(3):182–
185.
16. Cheung LP, Li KC, Hollett MD, Bergman
AG, Herfkens RJ. Meniscal tears of the knee:
accuracy of detection with fast spin-echo MR
radiology.rsnajnls.org ▪ Radiology: Volume 252: Number 2—August 2009
MUSCULOSKELETAL IMAGING: Assessment of Knee Joint with FSE-Cube
imaging and arthroscopic correlation in 293
patients. Radiology 1997;203(2):508 –512.
17. Ha TP, Li KC, Beaulieu CF, et al. Anterior
cruciate ligament injury: fast spin-echo MR
imaging with arthroscopic correlation in 217
examinations. AJR Am J Roentgenol 1998;
170(5):1215–1219.
18. Schaefer FK, Schaefer PJ, Brossmann J,
et al. Value of fat-suppressed PD-weighted
TSE-sequences for detection of anterior and
posterior cruciate ligament lesions: comparison to arthroscopy. Eur J Radiol 2006;
58(3):411– 415.
19. Escobedo EM, Hunter JC, Zink-Brody GC,
Wilson AJ, Harrison SD, Fisher DJ. Usefulness of turbo spin-echo MR imaging in the
evaluation of meniscal tears: comparison
with a conventional spin-echo sequence. AJR
Am J Roentgenol 1996;167(5):1223–1227.
20. Schweitzer ME, Tran D, Deely DM, Hume
EL. Medial collateral ligament injuries: eval-
uation of multiple signs, prevalence and location of associated bone bruises, and assessment with MR imaging. Radiology 1995;
194(3):825– 829.
21. Mohr A, Roemer FW, Genant HK, Liess C.
Using fat-saturated proton density-weighted
MR imaging to evaluate articular cartilage
[letter]. AJR Am J Roentgenol 2003;181(1):
280 –281.
22. Mohr A. The value of water-excitation 3D
FLASH and fat-saturated PDw TSE MR imaging for detecting and grading articular cartilage lesions of the knee. Skeletal Radiol
2003;32(7):396 – 402.
23. Loredo R, Hodler J, Pedowitz R, Yeh LR,
Trudell D, Resnick D. Posteromedial corner
of the knee: MR imaging with gross anatomic
correlation. Skeletal Radiol 1999;28(6):305–
311.
24. Yu JS, Salonen DC, Hodler J, Haghighi P,
Trudell D, Resnick D. Posterolateral aspect
Radiology: Volume 252: Number 2—August 2009 ▪ radiology.rsnajnls.org
Kijowski et al
of the knee: improved MR imaging with a
coronal oblique technique. Radiology 1996;
198(1):199 –204.
25. Rajeswaran G, Lee JC, Healy JC. MRI of the
popliteofibular ligament: isotropic 3D WEDESS versus coronal oblique fat-suppressed
T2W MRI. Skeletal Radiol 2007;36(12):
1141–1146.
26. Blackmon GB, Major NM, Helms CA. Comparison of fast spin-echo versus conventional
spin-echo MRI for evaluating meniscal tears.
AJR Am J Roentgenol 2005;184(6):1740 –
1743.
27. Rubin DA, Kneeland JB, Listerud J, UnderbergDavis SJ, Dalinka MK. MR diagnosis of meniscal
tears of the knee: value of fast spin-echo vs conventional spin-echo pulse sequences. AJR Am J
Roentgenol 1994;162(5):1131–1135.
28. Stevens KJ, Busse RF, Han E, et al. Ankle:
isotropic MR imaging with 3D-FSE-Cube—
initial experience in healthy volunteers. Radiology 2008;249(3):1026 –1033.
495