Automatic Edge Detection in Echocardiographic Doppler Images Using an Improved Canny Algorithm.

This paper introduces an optimized edge detection algorithm for the detection of tricuspid regurgitant jet velocity (TRV) trace edges in echocardiographic Doppler images of patients screened for pulmonary hypertension (PH). The proposed algorithm combines mathematical morphology with the Canny algorithm to overcome the limitations of traditional edge detection methods. A cohort of 348 patients was used for evaluation through comparing the optimized algorithm with manual labeling and the standard Canny algorithm. The results demonstrate the superiority of the optimized algorithm in accurately detecting TRV trace edges, with improved completeness, correctness, and overall quality metrics. The clinical implications of this algorithm include enhanced diagnosis, streamlined workflow, and improved patient prognosis. The algorithm provides a reliable and efficient tool for TRV trace edges detection for PH screening.