The comparison of two approaches for detecting and locating abnormalities on coronary artery images.

Coronary artery disease (CAD) is believed to be one of the most harmful fatal diseases in the world. An experienced doctor needs a lot of time to diagnose CAD in a patient. We proposed two methods to detect abnormal positions from coronary artery imaging to improve efficiency and performance in diagnostic abnormalities. In the first method, we introduce the vessel wall browsing algorithm to locate abnormalities on the blood vessels by comparing the distances from baseline to points under consideration. This algorithm reached 71.4%. We apply a convolutional neural network (CNN) model to predict whether a coronary image is normal or abnormal in the second method. The result from the experiment using our private dataset shows that our methods have an accuracy of 67.7%.