Pre-operative evaluation of patients undergoing knee articular cartilage repair: MRI 3D thickness maps derived from a validated, automated segmentation platform - initial results

Abstract

s / Osteoarthritis and Cartilage 21 (2013) S63–S312 S202 disease using backscattered electron imaging of embedded tissue blocks. These structures are formed by the extrusion of mineralisable matrix through microcracks in the articular calcified cartilage. Recently using micro-computed tomography, we identified similar structures in a human femoral head obtained from a patient with osteoarthropathy of alkaptonuria who underwent arthroplasty to alleviate severe joint pain. These protrusions are more densely mineralised than bone and mineralised cartilage. They protrude so far within articular cartilage that they are at risk of breaking off to generate small, hard, sharp edged fragments which might act as abrasive cutting and grinding particles that could contribute to the mechanical destruction of cartilage. The aim of this study was to determine if these protrusions could be identified by MRI. Methods: Femoral head and neck samples were fixed in formalin solution and then transferred to 70% ethanol. They were scanned in a Trio 3 Tesla MRI scanner using a wrist coil. Isotropic voxel data were acquired using a range of different modalities including dual echo steady state (DESS) sequences. Results: We found that the hypermineralised protrusions were visible, silhouetted against the signal-rich articular cartilage on the DESS sequences at resolutions of 0.5mm (TR1⁄416.3ms; TE1⁄44.7ms; NEX1⁄41; 320x320x160 voxels) and 0.23mm (TR1⁄418.2; TE1⁄45.64ms; NEX1⁄45; 512x512x240 voxels). Using MRI at both these resolutions, wewere able to detect protrusions similar to those previously identified by microtomography in the femoral head of the patient with alkaptonuria. Furthermore we were able to detect the protrusions in a femoral head from a patient with osteoarthritis unrelated to alkaptonuria. Conclusions: The results of this study demonstrate that hypermineralised subchondral protrusions are present in human femoral heads in the rare disease alkaptonuria but also in osteoarthritis. Although the MR images showing protrusions were recorded postoperatively from femoral heads, the resolution at which they could be detected raises the prospect of finding them and monitoring their progression in situ with MRI. 382 PRE-OPERATIVE EVALUATION OF PATIENTS UNDERGOING KNEE ARTICULAR CARTILAGE REPAIR: MRI 3D THICKNESS MAPS DERIVED FROM A VALIDATED, AUTOMATED SEGMENTATION PLATFORM INITIAL RESULTS J.M. Farber y, J. Tamez-Pena y, S. Totterman y, J. Larkin z, B. Holladay z, F. Heis z. yQmetrics, Rochester, NY, USA; zCommonwealth Orthopaedic Ctr.s, Edgewood, KY, USA Purpose: To present a robust, automatedMR imagingmethodology that generates 3D articular cartilage (AC) thickness maps of the knee, and which delineates the size and location of AC defects, as well as the thickness and integrity of the defect walls, to serve as an accurate preoperative guide for AC defect repair. Methods: A sagittal 3D FSE FS sequence (TR-2300; TE-20) is obtained on all patients undergoing routine 1.5TMRI (Optima 450W, GE,Milwaukee) of the knee at our facility. These 3D FSE data sets are sent to a dedicated work station (Qmetrics Technologies, Rochester, NY), which automatically segments the knee AC. The segmented images are then reviewed bya radiologist for accurate fit with the source images. If necessary, edits can be made on the dedicated work station to ensure proper segmentation. Themethodology of this process has been validated (1). From the segmented data sets, 3DAC thicknessmaps are generated automatically. These thickness maps are then reviewed by the radiologist for accuracy, compared to the source images. In patients who underwent subsequent AC defect repair, intra operative measurements were obtained of the AC defect size and location. The intra operative, gold standard data were then compared to the prospectively obtained thickness maps to assess accuracy of lesion size and location. Results: Initial results (N1⁄412) validate the accuracy of the thickness maps in delineating AC defect size and location (Fig. 1). In addition, the thickness maps accurately delineated the integrity of the AC defect walls, allowing for accurate surgical debridement and pre-operative planning for graft material allocation. In no case did the thickness map miss an AC defect of the segmented region. Conclusions: In our initial experience, 3D thickness maps of knee AC defects accurately detect and delineate AC defects, and are a clinically useful pre-operative tool. In addition, these automatically generated maps, which can be displayed along any axis, are useful in patientphysician discussions when discussing treatment options. Further work, specifically testing a larger N, is required to obtain statistical validation. Further work also will include also the use of T2 maps, with the thickness maps as a template (Image Available), to further assess the health of the AC surrounding a defect (2), to optimize further intra operative wall debridement and graft material allocation. Ă