Automatic System for Diseased Artery Transverse Section Detection

This article deals with an artery transversal section detection from ultrasound medical images. Exact localization of artery is needed for measuring artery properties, which can be used for diagnosing vascular diseases such as atherosclerosis, hypoplasia. Image processing methods are non-invasive, fast and cheap. In this work, the Viola-Jones object detector was used and further adopted for artery detection problem. In comparison with the other approaches created method improves speed and accuracy of detection mainly for unhealthy arteries. Created methods achieves 96% accuracy for healthy arteries and 95% accuracy of detection for unhealthy arteries. Average detection time per image was 2.5 milliseconds.