관상동맥 혈관내부 초음파 영상에서 내벽 및 외벽 윤곽선 자동추출을 위한 영상처리 알고리즘 개발

Development of an image processing system to detect automatically intimal and adventitial contours from intravascular ultrasound images

Intravascular ultrasound images of coranary artery contain very important informations on heart disease. The intimal contours on the image show informations and data to examine intravascular problems of patients. A new computation algorithm to detect the intimal and adventitial contours from the intravascular images was developed. An Image processing on gray level image was used. It uses arrays of pixels in each radial lines on the images. A "Robert" filter was adopted at first step for one dimensional image processing. Some other calculation techniques were developed to inclose the accuracy of automatically detected contours. The standard contour data to compare with automatically detected contour data were obtained through manually tracing by experienced cardiological medical doctors. The result of the new algorithm shows high accuracy of 80 % matching with the standard contour data.