Computer-Aided Diagnosis Scheme for Detecting Significant Stenosis on Whole Heart Coronary MR Angiography Based on Signal Intensity and Luminal Diameter

Ryohei Nakayama¹, Masaki Ishida¹, Nobuo Nakako¹, Motonori Nagata¹, Kakuya Kitagawa¹, Hajime Sakuma¹

A computer-aided diagnosis (CAD) was developed for detecting significant stenosis on whole heart coronary MRA. Coronary arterial segments were automatically classified according to AHA model. At each voxel on the centerline, a classifier based on the Mahalanobis distance for the changes of signal intensity and luminal diameter was employed for distinguishing between arteries with and without stenosis. In 34 patients with suspected coronary artery disease, the sensitivity and specificity of the CAD were 80% and 81%. The CAD scheme developed in this study showed high diagnostic performance and will have a substantial impact on interpretation of coronary MRA.