A Sequential Boundary Detection Algorithm for 3D Reconstruction of Ultrasound Image

• Main Authors: Ku, Bowen, 古博文
• Other Authors: Jenho Tsao
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/28781350616386106546

碩士 === 國立臺灣大學 === 電機工程學系研究所 === 86 === Although 3D ultrasound imaging is nowadays providing many successful applications, particularly in displaying fetal and vascular anatomy, it is still difficult to do boundary detection accurately for other applications, which results in incorrect shape of target tissue. We propose an algorithm which can detect the boundary of target tissue from sequentially scanned images. The algorithm is implemented in seven steps: (1) operator initialization; (2) corners linking; (3) correction of the initial boundary; (4) correction smoothing; (5) boundary linking; (6) linkage checking; and (7) corners searching then step 2 for incoming image. To keep the linkage of the boundary, our strategy is polygon approximation. We also emphasize the utilization of temporal and spatial information. We give some simulation and clinical experience to check the performance of the algorithm, and get good results.