Rapid volumetric mri measures of the medial meniscus for osteoarthritis studies: validity and responsiveness of a new automated method

Purpose: Position, integrity and volume (hypertrophy and maceration) of the medial meniscus may play a pivotal role in osteoarthritis (OA) development and progression. There have been few reports of quantitative tools to assess meniscal changes; those that exist generally reflect 2-dimensial measures and do not use automated software. To address this we developed and validated a semi-automated software method to rapidly characterize the meniscus volume on knee MRI scans. Methods: Forty-one subjects from the Osteoarthritis Initiative (OAI) were assessed. Thirty subjects were randomly selected from the Control Cohort (no radiographic knee OA) at baseline and 12 months. An additional 11 subjects from the Incidence Cohort were assesed, also free of radiographic OA at baseline but with known meniscus maceration (based on semiquantitative MOAKS scoring) or meniscus body extrusion (from manual quantitative measures). The 11 scans were evaluated for volume changes from the baseline to 48-month time points. The scans were analyzed using a new semi-automated software tool on the same sequence (Coronal 3D T1-weighted Fast Low Angle Shot Water Excitation (FLASH WE) (0.313 x 0.313 x 1.5 mm, TR 20 ms, TE 7.5 ms)), obtained on a 3-T Siemens Trio MR system. Meniscal body volume and extrusions measures were taken from the central 5 slices. The central slice (+/−2 slices) was halfway between the anterior and posterior tibial limits (first/last image in which the tibial cartilages were identified). Semi-automated image analysis software was used to delineate the meniscus margins on each of the five slices. Segmentation was initiated by the placement of a seed point over the meniscus on the center slice. The software then attempted to outline the margins of the medial compartment meniscus on each of the five slices. Tools to manually correct the computer-generated contours were also used. Twenty Control Cohort scans were read at baseline and follow-up for repositioning reliability by one reader (CR), blinded to time point. Ten separate baseline scans were read two times by the same reader for intra-rater reliability. A second reader (JD) also read these scans for inter-rater reliability. Both readers were blinded to each others readings, subject identifiers, and previous readings. For the 11 subjects with known meniscal changes, the extrusion ratio (proportion of the total medial meniscus volume located external to the medial edge of the tibial plateau) was determined, depicted by the red line in Figure 1. Descriptive statistics for the sample, total meniscal body volume, extrusion ratio and reader times were calculated. ICC’s and 95% CI (2-way mixed model - subjects random, raters fixed) were determined for the reliability of patient repositioning by comparing the baseline and 12 month scans, and for inter and intra reader reliability. Responsiveness for the extrusion metric was assessed using the standardized response means (SRMs) comparing baseline to 48 month extrusion volume fraction Results: The mean age of the subjects at baseline was 54.4 years, 24 (59%) were women, and 35 (85%) were Caucasian. All knees had baseline Kellgren and Lawrence scores of 0 or 1. Average baseline and 12 mo Control Cohort menuscus volumes (SD) were 131.0 (37.0) mmˆ3 and 130.3 (26.6) mmˆ3. The average (SD) baseline and 48mo meniscus extrusinon ratios for the 11 knes from the Incidence cohortwere 0.43 (0.16) and 0.50 (0.21). The ICCs (95% CI) for meniscal volume were as follows: Intra-rater Reliablity 0.95 (0.79–0.99), Inter-rater Reliability: 0.90 (0.58, 0.97) Re-positioning Reliability: 0.86 (0.63, 0.94). The SRM value for responsiveness was −0.31. The reader time was approximately 3 minutes per scan. Conclusions: We provide evidence that a new rapid semi-automated software method to quantify medial meniscal body volume is responsive and reproducible. This method is potentially practical for use in large longitudinal cohort studies of knee OA.