UC San Diego

4D cardiac CT is increasingly used to evaluate cardiac dynamics. Echocardiography and CMR have demonstrated the utility of longitudinal shortening (LS) measures. We demonstrate the ability of a recently published deep learning framework to automatically and accurately measure LS from CT for detection of wall motion abnormalities (WMA) and Mitral Annular Plane Systolic Excursion (MAPSE).100 clinical cineCT studies were evaluated by three experienced cardiac CT readers for presence of WMA: 50 for method development and 50 for testing. Previously developed convolutional neural network was used to automatically segment the LV bloodpool and to define the 2CH, 3CH, and 4CH long-axis imaging planes. LS was measured as the perimeter of the bloodpool for each long-axis plane. Two smoothing approaches were developed to avoid artifacts. The impact of the smoothing was evaluated by comparing LS estimates to LV ejection fraction and the fractional area change of the corresponding view. Our approach successfully analyzed 48/50 patients in the training cohort and 47/50 in the testing cohort. Smoothing significantly improved agreement between LS and fractional area change (R2: 2CH=0.38 vs 0.88 vs 0.92). The optimal single LS cutoff for identification of WMA in all LAX views was -17.0% in the training cohort. This led to correct labeling of 85% of the views in the testing cohort. Per-study accuracy was 83% (79% sensitivity and 86% specificity). LS values accurately identify regional wall motion abnormalities and may be used to complement standard visual assessments. MAPSE showed less utility in classification of WMA compared with LS.