Evaluation of Edge Detection and Fusion Methods for Angiographic Image Processing

Abstract

This study explores the application of digital image processing techniques to improve the diagnostic accuracy of coronary artery disease through enhanced analysis of angiographic images. A curated dataset of 40 grayscale, single-vessel angiograms was used to evaluate the effectiveness of various preprocessing and edge detection methods. Preprocessing steps included noise reduction, histogram equalization, and morphological operations aimed at improving image clarity and highlighting vascular structures. Three edge detection algorithms, Otsu, Canny, and Roberts, were applied, and their outputs were further combined using the Dempster-Shafer fusion theory. Performance was assessed using edge-based structural similarity (ESSIM) and the percentage error in vessel diameter measurements. The Canny algorithm demonstrated the highest individual accuracy, while the fusion of Canny and Roberts yielded superior results, achieving the lowest error and the highest similarity index.