Comparison of cone-beam and fan-beam computed tomography and low-field magnetic resonance imaging for detection of palmar/plantar osteochondral disease in Thoroughbred horses.

Abstract

BACKGROUND: Palmar/plantar osteochondral disease (POD) of the metacarpal/tarsal condyles is a common pathological finding in racehorses. OBJECTIVE: To compare diagnoses, imaging details, and measurements of POD lesions between cone-beam computed tomography CT (CBCT), fan-beam CT (FBCT), and low-field magnetic resonance imaging (MRI) using macroscopic pathology as a gold standard. STUDY DESIGN: Cross-sectional study. METHODS: Thirty-five cadaver limbs from 10 horses underwent CBCT, FBCT, MRI, and macroscopic examination. CT and MR images were examined for presence of POD, imaging details of POD, and measurements of POD dimensions and areas. Imaging diagnoses, details, and measurements were compared with macroscopic examination and between modalities. RESULTS: Forty-eight POD lesions were seen over 70 condyles. Compared with macroscopic examination the sensitivity and specificity of diagnosis were 95.8% (CI95 = 88%-99%) and 63.6% (CI95 = 43%-81%) for FBCT, 85.4% (CI95 = 74%-94%) and 81.8% (CI95 = 63%-94%) for CBCT, and 69.0% (CI95 = 54%-82%) and 71.4% (CI95 = 46%-90%) for MRI. Inter-modality agreement on diagnosis was moderate between CBCT and FBCT (κ = 0.56, p < 0.001). POD was identified on CT as hypoattenuating lesions with surrounding hyperattenuation and on MRI as either T1W, T2W, T2W, and STIR hyperintense lesions or T1W and T2W heterogeneous hypointense lesions with surrounding hypointensity. Agreement on imaging details between CBCT and FBCT was substantial for subchondral irregularity (κ = 0.61, p < 0.001). Macroscopic POD width strongly correlated with MRI (r = 0.81, p < 0.001) and CBCT (r = 0.79, p < 0.001) and moderately correlated with FBCT (r = 0.69, p < 0.001). Macroscopic POD width was greater than all imaging modality (p < 0.001). MAIN LIMITATIONS: Effect of motion artefact in live horse imaging could not be assessed. CONCLUSIONS: All imaging modalities were able to detect POD lesions, but underestimated lesion size. The CT systems were more sensitive, but the differing patterns of signal intensity may suggest that MRI can detect changes associated with POD pathological status or severity. The image features observed by CBCT and FBCT were similar.