AI-Enhanced High-Precision Segmentation and Perfusion Analysis in Myocardial Contrast Echocardiography

Myocardial contrast echocardiography (MCE) facilitates the quantification of myocardial perfusion and aids in diagnosing coronary artery disease (CAD). However, noise and artifacts in MCE data often hinder accurate perfusion analysis. This study presents an artificial intelligence-enhanced method for addressing these challenges. The method divides perfusion quantification into three steps: segmentation, segmental division, and parameter extraction. This method includes an AI model specifically designed for myocardial segmentation in MCE data. Evaluated on a custom dataset from Fuwai Hospital, the AI model achieved a Dice coefficient of 0.88 and an intersection over union (IoU) of 0.78 for the apical four-chamber (A4C) view, outperforming existing models. The segmented myocardium was divided into seven regions to extract perfusion parameters, with cardiac pose features across multiple cardiaccycles incorporated to enhance accuracy. By automating and streamlining perfusion assessment, the proposed method demonstrates potential to improve the clinical application of MCE in computer-aided diagnosis.