Utility of the Heart-to-Single Vertebra Ratio Compared with Artificial Intelligence-Derived Vertebral Heart Score for Detecting Cardiomegaly in Dogs with Myxomatous Mitral Valve Degeneration

The heart-to-single vertebra ratio measured at the seventh thoracic vertebra (HSVR-T7) is a recently described radiographic method used to assess heart size in dogs. Our study evaluated the diagnostic performance of HSVR-T7 for identifying echocardiographically defined cardiomegaly in dogs with myxomatous mitral valve disease (MMVD). Its correlation with vertebral heart score (VHS) and vertebral left atrial size (VLAS), and as its discriminatory performance compared with artificial intelligence (AI)-derived VHS and VLAS measurements were investigated. Radiographs of 93 dogs with echocardiographically confirmed myxomatous mitral valve disease were included; 35 were classified as ACVIM stage B1 and 58 as stage B2 or C. Among the 35 stage B1 dogs, five (14.3%) had isolated left atrial enlargement and two (5.7%) had isolated left ventricular enlargement. The HSVR-T7 scores were significantly higher in dogs with echocardiographically confirmed cardiomegaly (stage B2 or C) than in dogs without cardiomegaly (stage B1): 11.65 ± 1.13 versus 10.47 ± 1.08 (p < 0.001). At the optimal cut-off value of ≥11.15, HSVR-T7 yielded an area under the curve (AUC) of 0.769 of the receiver operating characteristic (ROC) curve, with 70.7% sensitivity and 74.3% specificity. The HSVR-T7 showed a very strong positive correlation with VHS (r = 0.890; p < 0.001). Artificial intelligence-derived VHS showed the highest diagnostic performance at an optimal cut-off value of ≥11.35 (AUC, 0.861; sensitivity, 75.9%; specificity, 82.9%), whereas AI-derived VLAS yielded an AUC of 0.765, with 62.1% sensitivity and 85.7% specificity at an optimal cut-off value of ≥2.65. Overall, HSVR-T7 was less sensitive and specific in detecting cardiomegaly in dogs with myxomatous mitral valve disease than VHS, and it cannot replace echocardiographic, VHS or VLAS measurements.