Semi-Automated Image Segmentation for Radiotherapy Treatment Planning in Lung Cancer

• Main Authors: Jen-Hung Chi, 紀任鴻
• Other Authors: 程大川
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/23pkw7

碩士 === 中國醫藥大學 === 生物醫學影像暨放射科學學系碩士班 === 105 === Radiotherapy is currently one of the main treatment of the cancer. In order to avoid the normal tissue to accept too much radiation, the radiotherapy treatment planning will be designed before clinical radiation therapy. The Radiotherapy treatment planning is usually performed by experienced radiophysicists and oncologists, who are according to personal experience and imaging information, routinely contour organs near the tumor and the treatment range, and this process usually spends a lot of time. We believe that the contours of the organ can be obtained by the result of image segmentation, and if reducing the time of the contouring process during designing the radiotherapy treatment planning can increase the efficiency and reduce the human consumption. We used the random walks algorithm to segment the image, and built a graphical user interface system to assist the user to set seed points, using the random walks algorithm for image segmentation obtained organ contours through the seed points are provided by the user. We contoured the lungs, trachea, heart, spinal cord, body and target tumor volume in the experiment, and compared with the contours of the experts. In our experimental results, the most of the results of each organ in contour cases and clinical cases were good, only slightly higher in the volume of the spinal cord. In the results of the calculus, the average amount of time spent in each study in our study was about two minutes, and it was effectively to save time.