Application of a Globally Optimal 3D Iterative Closest Point Point-Set Registration in CT Angiography

Chin-Wen Lin ； Jui-Chiu Chiang ； Shao-Jer Chen ； Po-Yu Huang ； Kai-Xiang Lin ； You-Jun Chen ； Chih-Wen Lin ； Hou-Chi Hwang

《Journal of Radiological Science》 46卷 (2021/03) Pp. 39-43

https://doi.org/10.6698/JRS.202103_46.0009

摘要

PURPOSE. In clinical practice, three-dimensional (3D) registration of medical images is always a challenge because of the movement of the patient. A 3D registration method was applied in this study to extract computed tomography angiography (CTA) images. MATERIALS AND METHODS. A study of 26 slab regions with various ranges of motion from 13 patients (10 men, 3 women; age range, 48-89 years) was approved by our institutional review board and evaluated retrospectively. Noncontrast-enhanced CT (NECT) and CTA images of the same volume data were acquired. Following simple thresholding, centralization, and normalization of the NECT and CTA images, a globally optimized 3D Iterative Closest Point (ICP) point-set registration was used for bone subtraction. Two radiologists conducted clinical evaluations, and the resulting data were analyzed. RESULTS. Significant improvements in motion correction and bony alignment were observed in the NECT and CTA images after the registration (p ＜ 0.001). The resultant images exhibited vascular extraction with bone and calcification elimination. CONCLUSION. A 3D ICP registration method can successfully extract the vascular structures from NECT and CTA images. The method is applicable in a wide variety of medicine and engineering contexts.

關鍵字

computed tomography angiography ； three-dimensional image registration ； blood vessel image segmentation