Computer-aided quantification of focal cartilage lesions using MRI: accuracy and initial arthroscopic comparison

Accuracy of cartilage-specific 3-Tesla 3D-DESS magnetic resonance imaging in the diagnosis of chondral lesions: comparison with knee arthroscopy

Background

Arthroscopy is considered as “the gold standard” for the diagnosis of traumatic intraarticular knee lesions. However, recent developments in magnetic resonance imaging (MRI) now offer good opportunities for the indirect assessment of the integrity and structural changes of the knee articular cartilage. The study was to investigate whether cartilage-specific sequences on a 3-Tesla MRI provide accurate assessment for the detection of cartilage defects.

Methods

A 3-Tesla (3-T) MRI combined with three-dimensional double-echo steady-state (3D-DESS) cartilage specific sequences was performed on 210 patients with knee pain prior to knee arthroscopy. Sensitivity, specificity, and positive and negative predictive values of magnetic resonance imaging were calculated and correlated to the arthroscopic findings of cartilaginous lesions. Lesions were classified using the modified Outerbridge classification.

Results

For the 210 patients (1260 cartilage surfaces: patella, trochlea, medial femoral condyle, medial tibia, lateral femoral condyle, lateral tibia) evaluated, the sensitivities, specificities, positive predictive values, and negative predictive values of 3-T MRI were 83.3, 99.8, 84.4, and 99.8 %, respectively, for the detection of grade IV lesions; 74.1, 99.6, 85.2, and 99.3 %, respectively, for grade III lesions; 67.9, 99.2, 76.6, and 98.2 %, respectively, for grade II lesions; and 8.8, 99.5, 80, and 92 %, respectively, for grade I lesions.

Conclusions

For grade III and IV lesions, 3-T MRI combined with 3D-DESS cartilage-specific sequences represents an accurate diagnostic tool. For grade II lesions, the technique demonstrates moderate sensitivity, while for grade I lesions, the sensitivity is limited to provide reliable diagnosis compared to knee arthroscopy.

Accuracy of cartilage-specific 3-Tesla 3D-DESS magnetic resonance imaging in the diagnosis of chondral lesions: comparison with knee arthroscopy

BACKGROUND

Arthroscopy is considered as "the gold standard" for the diagnosis of traumatic intraarticular knee lesions. However, recent developments in magnetic resonance imaging (MRI) now offer good opportunities for the indirect assessment of the integrity and structural changes of the knee articular cartilage. The study was to investigate whether cartilage-specific sequences on a 3-Tesla MRI provide accurate assessment for the detection of cartilage defects.

METHODS

A 3-Tesla (3-T) MRI combined with three-dimensional double-echo steady-state (3D-DESS) cartilage specific sequences was performed on 210 patients with knee pain prior to knee arthroscopy. Sensitivity, specificity, and positive and negative predictive values of magnetic resonance imaging were calculated and correlated to the arthroscopic findings of cartilaginous lesions. Lesions were classified using the modified Outerbridge classification.

RESULTS

For the 210 patients (1260 cartilage surfaces: patella, trochlea, medial femoral condyle, medial tibia, lateral femoral condyle, lateral tibia) evaluated, the sensitivities, specificities, positive predictive values, and negative predictive values of 3-T MRI were 83.3, 99.8, 84.4, and 99.8 %, respectively, for the detection of grade IV lesions; 74.1, 99.6, 85.2, and 99.3 %, respectively, for grade III lesions; 67.9, 99.2, 76.6, and 98.2 %, respectively, for grade II lesions; and 8.8, 99.5, 80, and 92 %, respectively, for grade I lesions.

CONCLUSIONS

For grade III and IV lesions, 3-T MRI combined with 3D-DESS cartilage-specific sequences represents an accurate diagnostic tool. For grade II lesions, the technique demonstrates moderate sensitivity, while for grade I lesions, the sensitivity is limited to provide reliable diagnosis compared to knee arthroscopy.

Mapping tibiofemoral gonarthrosis: an MRI analysis of non-traumatic knee cartilage defects

OBJECTIVE

Arthroscopy is "the gold standard" for the diagnosis of knee cartilage lesions. However, it is invasive and expensive, and displays all the potential complications of an open surgical procedure. Ultra-high-field MRI now offers good opportunities for the indirect assessment of the integrity and structural changes of joint cartilage of the knee. The goal of the present study is to determine the site of early cartilaginous lesions in adults with non-traumatic knee pain.

METHODS

3-T MRI examinations of 200 asymptomatic knees with standard and three-dimensional double-echo steady-state (3D-DESS) cartilage-specific sequences were prospectively studied for early degenerative lesions of the tibiofemoral joint. Lesions were classified and mapped using the modified Outerbridge and modified International Cartilage Repair Society classifications.

RESULTS

A total of 1437 lesions were detected: 56.1% grade I, 33.5% grade II, 7.2% grade III and 3.3% grade IV. Cartographically, grade I lesions were most common in the anteromedial tibial areas; grade II lesions in the anteromedial L5 femoral areas; and grade III in the centromedial M2 femoral areas.

CONCLUSION

3-T MRI with standard and 3D-DESS cartilage-specific sequences demonstrated that areas predisposed to early osteoarthritis are the central, lateral and ventromedial tibial plateau, as well as the central and medial femoral condyle.

ADVANCES IN KNOWLEDGE

In contrast with previous studies reporting early cartilaginous lesions in the medial tibial compartment and/or in the medial femoral condyle, this study demonstrates that, regardless of grade, lesions preferentially occur at the L5 and M4 tibial and L5 and L2 femoral areas of the knee joint.

In vitro assessment of knee MRI in the presence of metal implants comparing MAVRIC-SL and conventional fast spin echo sequences at 1.5 and 3 T field strength

PURPOSE

To assess lesion detection and artifact size reduction of a multiacquisition variable-resonance image combination, slice encoding for metal artifact correction (MAVRIC-SEMAC) hybrid sequence (MAVRIC-SL) compared to standard sequences at 1.5T and 3T in porcine knee specimens with metal hardware.

MATERIALS AND METHODS

Artificial cartilage and bone lesions of defined size were created in the proximity of titanium and steel screws with 2.5 mm diameter in 12 porcine knee specimens and were imaged at 1.5T and 3T magnetic resonance imaging (MRI) with MAVRIC-SL PD and short T1 inversion recovery (STIR), standard fast spin echo (FSE) T2 PD, and STIR and fat-saturated T2 FSE sequences. Three radiologists blinded to the lesion locations assessed lesion detection rates on randomized images for each sequence using receiver operating characteristic (ROC). Artifact length and width were measured.

RESULTS

Metal artifact sizes were largest in the presence of steel screws at 3T (FSE T2 FS: 28.7 cm(2) ) and 1.5T (16.03 cm(2) ). MAVRIC-SL PD and STIR reduced artifact sizes at both 3T (1.43 cm(2) ; 2.46 cm(2) ) and 1.5T (1.16 cm(2) ; 1.59 cm(2) ) compared to FS T2 FSE sequences (27.57 cm(2) ; 13.20 cm(2) ). At 3T, ROC-derived AUC values using MAVRIC-SL sequences were significantly higher compared to standard sequences (MAVRIC-PD: 0.87, versus FSE-T2 -FS: 0.73 [P = 0.025]; MAVRIC-STIR: 0.9 vs. T2 -STIR: 0.78 [P = 0.001] and vs. FSE-T2 -FS: 0.73 [P = 0.026]). Similar values were observed at 1.5T. Comparison of 3T and 1.5T showed no significant differences (MAVRIC-SL PD: P = 0.382; MAVRIC-SL STIR: P = 0.071).

CONCLUSION

MAVRIC-SL sequences provided superior lesion detection and reduced metal artifact size at both 1.5T and 3T compared to conventionally used FSE sequences. No significant disadvantage was found comparing MAVRIC-SL at 3T and 1.5T, although metal artifacts at 3T were larger. J. Magn. Reson. Imaging 2015;41:1291-1299. © 2014 Wiley Periodicals, Inc.

Spectroscopic measurement of cartilage thickness in arthroscopy: ex vivo validation in human knee condyles

PURPOSE

To evaluate the accuracy of articular cartilage thickness measurement when implementing a new technology based on spectroscopic measurement into an arthroscopic camera.

METHODS

Cartilage thickness was studied by ex vivo arthroscopy at a number of sites (N = 113) in human knee joint osteoarthritic femoral condyles and tibial plateaus, removed from 7 patients undergoing total knee replacement. The arthroscopic image spectral data at each site were used to estimate cartilage thickness. Arthroscopically derived thickness values were compared with reference cartilage thickness as measured by 3 different methods: needle penetration, spiral computed tomography scanning, and geometric measurement after sample slicing.

RESULTS

The lowest mean error (0.28 to 0.30 mm) in the regression between arthroscopic and reference cartilage thickness was seen for reference cartilage thickness less than 1.5 mm. Corresponding values for cartilage thickness less than 2.0 and 2.5 mm were 0.32 to 0.40 mm and 0.37 to 0.47 mm, respectively. Cartilage thickness images--created by pixel-by-pixel regression model calculations applied to the arthroscopic images--were derived to demonstrate the clinical use of a camera implementation.

CONCLUSIONS

On the basis of this investigation on osteoarthritic material, when one is implementing the spectroscopic method for estimating cartilage thickness into an arthroscopic camera, errors in the range of 0.28 to 0.30 mm are expected. This implementation does not, however, influence the fact that the spectral method performs less well in the cartilage thickness region from 1.5 to 2.5 mm and cannot assess cartilage thicker than 2.5 mm.

CLINICAL RELEVANCE

Imaging cartilage thickness directly in the arthroscopic camera video stream could serve as an interesting image tool for in vivo cartilage quality assessment, in connection with cartilage diagnosis, repair, and follow-up.

A spectroscopic approach to imaging and quantification of cartilage lesions in human knee joints

We have previously described a technology based on diffuse reflectance of broadband light for measuring joint articular cartilage thickness, utilizing that optical absorption is different in cartilage and subchondral bone. This study is the first evaluation of the technology in human material. We also investigated the prospects of cartilage lesion imaging, with the specific aim of arthroscopic integration. Cartilage thickness was studied ex vivo in a number of sites (n = 87) on human knee joint condyles, removed from nine patients during total knee replacement surgery. A reflectance spectrum was taken at each site and the cartilage thickness was estimated using the blue, green, red and near-infrared regions of the spectrum, respectively. Estimated values were compared with reference cartilage thickness values (taken after sample slicing) using an exponential model. Two-dimensional Monte Carlo simulations were performed in a theoretical analysis of the experimental results. The reference cartilage thickness of the investigated sites was 1.60 ± 1.30 mm (mean ± SD) in the range 0-4.2 mm. Highest correlation coefficients were seen for the calculations based on the near-infrared region after normalization to the red region (r = 0.86) and for the green region (r = 0.80).

Non-invasive MRI assessment of the articular cartilage in clinical studies and experimental settings

Attrition and eventual loss of articular cartilage are important elements in the pathophysiology of osteoarthritis (OA). Preventing the breakdown of cartilage is believed to be critical to preserve the functional integrity of a joint. Chondral injuries are also common in the knee joint, and many patients benefit from cartilage repair. Magnetic resonance imaging (MRI) and advanced digital post-processing techniques have opened possibilities for in vivo analysis of cartilage morphology, structure, and function in healthy and diseased knee joints. Techniques of semi-quantitative scoring of human knee cartilage pathology and quantitative assessment of human cartilage have been developed. Cartilage thickness and volume have been quantified in humans as well as in small animals. MRI detected cartilage loss has been shown to be more sensitive than radiographs detecting joint space narrowing. It is possible to longitudinally study knee cartilage morphology with enough accuracy to follow the disease-caused changes and also evaluate the therapeutic effects of chondro-protective drugs. There are also several MRI methods that may allow evaluation of the glycosaminoglycan matrix or collagen network of articular cartilage, and may be more sensitive for the detection of early changes. The clinical relevance of these methods is being validated. With the development of new therapies for OA and cartilage injury, MR images will play an important role in the diagnosis, staging, and evaluation of the effectiveness of these therapies.

Correlation between arthroscopic and histopathological grading systems of articular cartilage lesions in knee osteoarthritis

PURPOSE

Arthroscopic and particularly histopathological assessments have been used to evaluate alterations of knee cartilage in osteoarthritis (OA). The aim of this study was to examine the correlation between an arthroscopic method to grade the severity of chondropathies and the histological/histochemical grading system (HHGS) applied to the corresponding articular cartilage areas in knee OA.

METHODS

The articular cartilage surface was examined by chondroscopy using the Beguin and Locker severity criteria, analysing the lesions in 72 chondroscopic areas. Afterwards, samples were obtained by dividing the cartilage surface of the medial tibiofemoral compartment of three OA knee joints into equal squares and they were evaluated histologically using the HHGS. The correlation between both grading methods was assessed using the weighted Kappa coefficient (K(w)).

RESULTS

The results obtained with both scores showed good agreement (K(w): mean+/-standard deviation, 0.619+/-0.071). While the average HHGS scores of the chondral samples showed a better agreement with arthroscopic grades 0, I and II, the arthroscopic evaluation has a tendency to overestimate chondral lesions for histological grades III and IV. The intra- and inter-observer reliability of the HHGS evaluation of chondral lesions was excellent (Intraclass Correlation Coefficient: 0.909 and 0.941, respectively).

CONCLUSION

In this study, we found a good quantitative correlation between established arthroscopic severity and histopathological scoring systems, particularly in less advanced lesions. Our results suggest that the arthroscopic method is a valuable tool in clinical research to score chondropathies in the medial femorotibial compartment of the OA knee, although some limitations should not be overlooked.

New techniques for cartilage magnetic resonance imaging relaxation time analysis: texture analysis of flattened cartilage and localized intra- and inter-subject comparisons

MR relaxation time measurements of knee cartilage have shown potential to characterize knee osteoarthritis (OA). In this work, techniques that allow localized intra- and inter-subject comparisons of cartilage relaxation times, as well as cartilage flattening for texture analysis parallel and perpendicular to the natural cartilage layers, are presented. The localized comparisons are based on the registration of bone structures and the assignment of relaxation time feature vectors to each point in the bone-cartilage interface. Cartilage flattening was accomplished with Bezier splines and warping, and texture analysis was performed with second-order texture measures using gray-level co-occurrence matrices (GLCM). In a cohort of five normal subjects the performance and reproducibility of the techniques were evaluated using T1rho maps of femoral knee cartilage. The feasibility of creating a mean cartilage relaxation time map is also presented. Successful localized intra- and inter-subject T1rho comparisons were obtained with reproducibility similar to that reported in the literature for regional T2. Improvement of the reproducibility of GLCM features was obtained by flattening the T1rho maps. The results indicate that the presented techniques have potential in longitudinal and population studies of knee OA at different stages of the disease.

In search of a gold standard of knee cartilage defect topographical documentation: "freehand" arthroscopic mapping and introduction of new concepts

Currently, a significant proportion of effort on cartilage tissue research is devoted to establishing the most effective method of inducing cartilage repair. However, the studies themselves lack uniform documentation and comparisons are difficult to make. If valuable therapeutic information is to be gained from future studies, it is important that an accurate system exists for assessing focal cartilage defects to allow comparison between studies and/or investigators.

Accuracy and reliability of MRI vs. laboratory measurements in an ex vivo porcine model of arthritic cartilage loss

PURPOSE

To quantify the accuracy of magnetic resonance imaging (MRI) measurement of change in cartilage volume due to thin linear excisions, simulating arthritic cartilage losses, by comparison with laboratory volume measurements in an ex vivo porcine model.

MATERIALS AND METHODS

We scanned 15 porcine patellae by T1-weighted spoiled gradient echo (SPGR) MRI at baseline and after excision of up to three thin layers of articular cartilage. Excised fragment volume was determined from density and weight. Postexcision scans were "fused" to the baseline scan by three-dimensional (3D) registration. This allowed automated recalculation of the remaining cartilage volume within a baseline region of interest (ROI) following each excision. We compared MRI estimates of change in cartilage volume to direct laboratory measurement of fragment volume.

RESULTS

Our 38 excised fragments averaged 0.16 mL, or approximately 7% of cartilage volume. MRI and laboratory estimates of total cartilage volume loss differed by 1.6% +/- 13.2% (mean, coefficient of variation [CV]). Accuracy was +/-0.1 mL for 95% of scans.

CONCLUSION

MRI estimates of small changes in porcine patellar cartilage volume were unbiased, reliable, and accurate to 0.1 mL. Despite a proportionately high error in the very thin fragments tested, achievement of similar accuracy in vivo would be adequate to detect approximately two years of osteoarthritic cartilage loss.

Quantitative MRI of cartilage and bone: degenerative changes in osteoarthritis

Magnetic resonance imaging (MRI) and quantitative image analysis technology has recently started to generate a great wealth of quantitative information on articular cartilage and bone physiology, pathophysiology and degenerative changes in osteoarthritis. This paper reviews semiquantitative scoring of changes of articular tissues (e.g. WORMS = whole-organ MRI scoring or KOSS = knee osteoarthritis scoring system), quantification of cartilage morphology (e.g. volume and thickness), quantitative measurements of cartilage composition (e.g. T2, T1rho, T1Gd = dGEMRIC index) and quantitative measurement of bone structure (e.g. app. BV/TV, app. TbTh, app. Tb.N, app. Tb.Sp) in osteoarthritis. For each of these fields we describe the hardware and MRI sequences available, the image analysis systems and techniques used to derive semiquantitative and quantitative parameters, the technical accuracy and precision of the measurements reported to date and current results from cross-sectional and longitudinal studies in osteoarthritis. Moreover, the paper summarizes studies that have compared MRI-based measurements with radiography and discusses future perspectives of quantitative MRI in osteoarthritis. In summary, the above methodologies show great promise for elucidating the pathophysiology of various tissues and identifying risk factors of osteoarthritis, for developing structure modifying drugs (DMOADs) and for combating osteoarthritis with new and better therapy.

Preliminary study on diffraction enhanced radiographic imaging for a canine model of cartilage damage

OBJECTIVE

To demonstrate the ability of a novel radiographic technique, Diffraction Enhanced Radiographic Imaging (DEI), to render high contrast images of canine knee joints for identification of cartilage lesions in situ.

METHODS

DEI was carried out at the X-15A beamline at Brookhaven National Laboratory on intact canine knee joints with varying levels of cartilage damage. Two independent observers graded the DE images for lesions and these grades were correlated to the gross morphological grade.

RESULTS

The correlation of gross visual grades with DEI grades for the 18 canine knee joints as determined by observer 1 (r2 = 0.8856, P = 0.001) and observer 2 (r2 = 0.8818, P = 0.001) was high. The overall weighted kappa value for inter-observer agreement was 0.93, thus considered high agreement.

CONCLUSION

The present study is the first study for the efficacy of DEI for cartilage lesions in an animal joint, from very early signs through erosion down to subchondral bone, representing the spectrum of cartilage changes occurring in human osteoarthritis (OA). Here we show that DEI allows the visualization of cartilage lesions in intact canine knee joints with good accuracy. Hence, DEI may be applicable for following joint degeneration in animal models of OA.

Magnetic resonance imaging (MRI) of articular cartilage in knee osteoarthritis (OA): morphological assessment

OBJECTIVE

Magnetic resonance imaging (MRI) is a three-dimensional imaging technique with unparalleled ability to evaluate articular cartilage. This report reviews the current status of morphological assessment of cartilage with quantitative MRI (qMRI), and its relevance for identifying disease status, and monitoring progression and treatment response in knee osteoarthritis (OA).

METHOD

An international panel of experts in MRI of knee OA, with direct experience in the analysis of cartilage morphology with qMRI, reviewed the existing published and unpublished data on the subject, and debated the findings at the OMERACT-OARSI Workshop on Imaging technologies (December 2002, Bethesda, MA) with scientists and clinicians from academia, the pharmaceutical industry and the regulatory agencies. This report reviews (1) MRI pulse sequence considerations for morphological analysis of articular cartilage; (2) techniques for segmenting cartilage; (3) semi-quantitative scoring of cartilage status; and (4) technical validity (accuracy), precision (reproducibility) and sensitivity to change of quantitative measures of cartilage morphology.

RESULTS

Semi-quantitative scores of cartilage status have been shown to display adequate reliability, specificity and sensitivity, and to detect lesion progression at reasonable observation periods (1-2 years). Quantitative assessment of cartilage morphology (qMRI), with fat-suppressed gradient echo sequences, and appropriate image analysis techniques, displays high accuracy and adequate precision (e.g., root-mean-square standard deviation medial tibia=61 microl) for cross-sectional and longitudinal studies in OA patients. Longitudinal studies suggest that changes of cartilage volume of the order of -4% to -6% occur per annum in OA in most knee compartments (e.g., -90 microl in medial tibia). Annual changes in cartilage volume exceed the precision errors and appear to be associated with clinical symptoms as well as with time to knee arthroplasty.

CONCLUSIONS

MRI provides reliable and quantitative data on cartilage status throughout most compartments of the knee, with robust acquisition protocols for multi-center trials now being available. MRI of cartilage has tremendous potential for large scale epidemiological studies of OA progression, and for clinical trials of treatment response to structure modifying OA drugs.

3.0 vs 1.5 T MRI in the detection of focal cartilage pathology--ROC analysis in an experimental model

OBJECTIVE

To use receiver operator characteristics (ROC) analysis for assessing the diagnostic performance of three cartilage-specific MR sequences at 1.5 and 3 T in detecting cartilage lesions created in porcine knees.

DESIGN

Eighty-four cartilage lesions were created in 27 porcine knee specimens at the patella, the medial and lateral femoral and the medial and lateral tibial cartilage. MR imaging was performed using a fat saturated spoiled gradient echo (SPGR) sequence (in plane spatial resolution/slice thickness: 0.20 x 0.39 mm2/1.5 mm) and two fat saturated proton density weighted (PDw) sequences (low spatial resolution: 0.31 x 0.47 mm2/3 mm and high spatial resolution: 0.20 x 0.26 mm2/2 mm). The images were independently analyzed by three radiologists concerning the absence or presence of lesions using a five-level confidence scale. Significances of the differences for the individual sequences were calculated based on comparisons of areas under ROC curves (A(Z)).

RESULTS

The highest A(Z)-values for all three radiologists were consistently obtained for the SPGR (A(Z) = 0.84) and the high-resolution (hr) PDw (A(Z) = 0.79) sequences at 3T. The corresponding A(Z)-values at 1.5 T were 0.77 and 0.69; the differences between 1.5 and 3 T were statistically significant (P < 0.05). A(Z)-values for the low-resolution PDw sequence were lower: 0.59 at 3 T and 0.55 at 1.5 T and the differences between 1.5 and 3T were not significant.

CONCLUSION

With optimized hr MR sequences diagnostic performance in detecting cartilage lesions was improved at 3 T. For a standard, lower spatial resolution PDw sequence no significant differences, however, were found.