An Improved Intelligent Scissors Algorithm for the Segmentation of Vessels Segments in Coronary Angiography Imaging

Interactive segmentation is widely used to precisely segment the parts of interest in medical images. Among them, the intelligent scissors algorithm is an efficient interactive segmentation algorithm, but it has poor performance in processing images with a high similarity between foreground and background. And it is easily affected by the strong edge, which leads to more seed points and longer time to segment the interested vessel segments in the coronary angiography images. In order to improve the segmentation efficiency, we redesign the cost function, using Canny operator, Scharr operator, to replace the Laplace operator in the traditional algorithm. Moreover, gradient magnitude adjustment function and histogram adjustment function are introduced to improve the segmentation effect. Experiment results show that the improved algorithm can shorten the segmentation time and improve the segmentation efficiency without the loss of performance compared with the traditional algorithm.