Ruoff CM, Eichelberger BM, Pool RR, et al.
Noninvasive identification of canine articular cartilage injuries is challenging. The objective of this prospective, cadaveric, diagnostic accuracy study was to determine if small field-of-view, three tesla magnetic resonance imaging (MRI) was an accurate method for identifying experimentally induced cartilage defects in canine stifle joints. Forty-two canine cadaveric stifles (n = 6/group) were treated with sham control, 0.5, 1.0, or 3.0 mm deep defects in the medial or lateral femoral condyle. Proton density-weighted, T1-weighted, fast-low angle shot, and T2 maps were generated in dorsal and sagittal planes. Defect location and size were independently determined by two evaluators and compared to histologic measurements. Accuracy of MRI was determined using concordance correlation coefficients. Defects were identified correctly in 98.8% (Evaluator 1) and 98.2% (Evaluator 2) of joints. Concordance correlation coefficients between MRI and histopathology were greater for defect depth (Evaluator 1: 0.68–0.84; Evaluator 2: 0.76–0.83) compared to width (Evaluator 1: 0.30–0.54; Evaluator 2: 0.48–0.68). However, MRI overestimated defect depth (histopathology: 1.65 ± 0.94 mm; Evaluator 1, range of means: 2.07–2.38 mm; Evaluator 2, range of means: 2–2.2 mm) and width (histopathology: 6.98 ± 1.32 mm; Evaluator 1, range of means: 8.33–8.8 mm; Evaluator 2, range of means: 6.64–7.16 mm). Using the paired t-test, the mean T2 relaxation time of cartilage defects was significantly greater than the mean T2 relaxation time of adjacent normal cartilage for both evaluators (P < 0.0001). Findings indicated that MRI is an accurate method for identifying cartilage defects in the cadaveric canine stifle. Additional studies are needed to determine the in vivo accuracy of this method.