AngioPy Segmentation Tool for Coronary Artery Analysis

This article presents a technical report on AngioPy, an open-source deep learning tool designed for coronary artery segmentation. The primary objective of AngioPy is to enhance the accuracy of quantitative coronary angiography (QCA) by minimizing the need for manual corrections typically required in traditional edge detection algorithms. The study utilizes a dataset of 2455 images from the FAME 2 trial and validates the model with an external dataset of 580 images. The performance metrics indicate a high average F1 score of 0.927, demonstrating the tool's effectiveness in achieving accurate segmentation results. The report details the methodology employed in developing the deep learning models, including the integration of user-defined ground-truth points to improve segmentation performance. Additionally, it discusses the implications of AngioPy for clinical practice, particularly in reducing subjectivity and enhancing the reliability of coronary assessments. The findings suggest that AngioPy has significant potential to improve the efficiency and accuracy of coronary imaging analysis.