Automatic Detection of Coronary Artery Stenosis in X-ray Angiograms using Gaussian Filters and Genetic Algorithms

摘要

This paper presents a new automatic coronary artery segmentation method based on Gaussian matched filters (GMF) and genetic algorithms in X-ray angiograms. This method is used for detecting coronary stenosis based on the second-order derivatives of the vessel size, and it consists of the steps of vessel enhancement and vessel segmentation. In the enhancement step, the proposed method uses genetic algorithms to obtain the best performance of the GMF by tuning the optimal parameters. The results of the detection method are compared with those obtained by different state-of-the-art methods based on GMF using the area (A_z) under the receiver operating characteristic curve. Moreover, in the segmentation step, the inter-class variance thresholding method has been applied for classifying vessel and non-vessel pixels. The segmentation results obtained by the proposed method demonstrated high performance achieving an accuracy of 0.962 with a test set of 20 angiograms of a database of 40 angiograms outlined by a specialist. In addition, the detection results of potential cases of coronary stenosis are promising and suggest that the proposed method can be highly suitable for computer-aided diagnosis in cardiology.