Mylène Jansen

Cartilage collagen structure after KJD and HTO using T2-mapping 305 15 Image acquisition Multi-slice multi-echo spin-echo (MSME) T2-mapping scans were performed on a clinical 3T MRI scanner (Achieva 3T; Philips Medical Systems) using a 16-channel knee coil. T2 relaxation times were obtained from T2 maps reconstructed using sagittal SE acquisition, with 8 echo times (TE) of 10, 20, 30, 40, 50, 60, 70, and 80 ms. The slice thickness was 3 mm, with a pixel matrix of 640x640 and a pixel size of 0.25x0.25 mm. In the same session, a sagittal proton-density weighted (PDW) scan with fat suppression was performed, with an echo time of 40 ms, slice thickness of 2.7 mm, pixel matrix of 528x528 mm and pixel size of 0.30x0.30 mm. Scans were performed before treatment (baseline) and at 1 years and 2 years after treatment. HTO patients did not undergo MRI scans at 1 year due to the metal-plate in situ . Only patients with scans available for analysis at baseline and 2-year follow-up were included in this study. Image analysis Segmentation was performed thrice for all images, by 3 independent observers (MJ, NB, CN). Based on initial experimental segmentation, a consensus was reached between the 3 observers on how to perform the segmentations. The knee joint was divided in 4 regions: lateral femur, medial femur, lateral tibia, and medial tibia. Segmentation began from the center of the joint and was performed on 7 slices, counting outwards from the first slice without cruciate ligaments. As was done for the dGEMRIC analyses in this same group of patients, regions reached until the most anterior part of the tibia plateau; the posterior tibial region reached until the most posterior part of the tibia plateau; and the posterior femoral regions encompassed all visible cartilage. 24 Regions of interests (ROI) were drawn on the PDW images using in- house developed software (Experimental Analysis, Image Sciences Institute) and automatically applied on the T2-mapping images, where manual corrections could be performed if necessary. From all scans, the volume (mm 3 ) and T2 relaxation times (ms) were calculated for each of the 4 segmented regions. Pixels with T2 relaxation times >100 ms were excluded, as these were not realistic for cartilage, but instead likely represented the bone edge included in the ROI. An example image with the included ROI is shown in Figure 1. Untreated control group The OAI used a comparable protocol with somewhat lower resolution for the acquisition of MSME MRIs. 26 Cartilage T2 times from the OAI were based on a quality-controlled manual segmentation of femorotibial cartilages and were used as an untreated group of OA patients. Cartilage T2 times were available at baseline, 1 year, and 4 years from previous analyses. 27,28 From the available subset of OAI knees with T2-mapping results, control patients were selected