Experimental model to evaluate in vivo and in vitro cartilage MR imaging by means of histological analyses

Abstract

Objectives Implementation of an experimental model to compare cartilage MR imaging by means of histological analyses. Material and methods MRI was obtained from 4 patients expecting total knee replacement at 1.5 and/or 3 T prior surgery. The timeframe between pre-op MRI and knee replacement was within two days. Resected cartilage-bone samples were tagged with Ethi ®-pins to reproduce the histological cutting course. Pre-operative scanning at 1.5 T included following parameters for fast low angle shot (FLASH: TR/TE/FA = 33 ms/6 ms/30°, BW = 110 kHz, 120 mm × 120 mm FOV, 256 × 256 matrix, 0.65 mm slice-thickness) and double echo steady state (DESS: TR/TE/FA = 23.7 ms/6.9 ms/40°, BW = 130 kHz, 120 × 120 mm FOV, 256 × 256 matrix, 0.65 mm slice-thickness). At 3 T, scan parameters were: FLASH (TR/TE/FA = 12.2 ms/5.1 ms/10°, BW = 130 kHz, 170 × 170 mm FOV, 320 × 320, 0.5 mm slice-thickness) and DESS (TR/TE/FA = 15.6 ms/4.5 ms/25°, BW = 200 kHz, 135 mm × 150 mm FOV, 288 × 320 matrix, 0.5 mm slice-thickness). Imaging of the specimens was done the same day at 1.5 T. MRI (Noyes) and histological (Mankin) score scales were correlated using the paired t-test. Sensitivity and specificity for the detection of different grades of cartilage degeneration were assessed. Inter-reader and intra-reader reliability was determined using Kappa analysis. Results Low correlation (sensitivity, specificity) was found for both sequences in normal to mild Mankin grades. Only moderate to severe changes were diagnosed with higher significance and specificity. The use of higher field-strengths was advantageous for both protocols with sensitivity values ranging from 13.6% to 93.3% (FLASH) and 20.5% to 96.2% (DESS). Kappa values ranged from 0.488 to 0.944. Conclusions Correlating MR images with continuous histological slices was feasible by using three-dimensional imaging, multi-planar-reformat and marker pins. The capability of diagnosing early cartilage changes with high accuracy could not be proven for both FLASH and DESS.