ISEKI Hiroshi
Department Research Institutes and Facilities, Research Institutes and Facilities Position |
|
Article types | Original article |
Language | Japanese |
Peer review | Peer reviewed |
Title | Marker-free augmented reality system for image-guided neurosurgery |
Domestic / Foregin | Domestic |
Volume, Issue, Page | pp.16-22 |
Publication date | 2012/02 |
Summary | In image-guided neurosurgery, augmented reality technology is adopted to overlay three-dimensionally reconstructed computer models of a target lesion onto a video image of the surgical field. Optical, electromagnetic, and mechanical tracking systems are usually used for registration between the reconstructed model and a view of the surgical field, but they are both expensive and not without technical problems. In this study, we propose a marker-free registration and tracking method. Initial registration is completed by manually selecting several pairs of anatomically-corresponding points on a reconstructed model and video image. Points are automatically detected (Harris corner detection algorithm) within a predefined region of interest, and a tracking algorithm for machine vision (KLT tracker) is applied for real-time target tracking. A feasibility study using a brain model showed that image overlay was achieved with refresh rate of 8 frames per second. Initial registration error was less than 0.5 mm when more than three pairs of anatomically-corresponding points were selected. This method is feasible and economically advantageous because it does not require an expensive tracking system. In a future study, we will evaluate feasibility and overlaying accuracy in the clinical setting. |
NAID | 110008802810 |