Combined CT-Video Registration and Tracking for Endoscopic Guidance
Open Access
- Author:
- Merritt, Scott Alexander
- Graduate Program:
- Electrical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- July 05, 2007
- Committee Members:
- William Evan Higgins, Committee Chair/Co-Chair
Jeffrey Louis Schiano, Committee Member
Shizhuo Yin, Committee Member
Robert T Collins, Committee Member - Keywords:
- Bronchoscopic Guidance
Endoscopic Guidance
Endoscope Tracking
Computer Vision
Image Registration
Virtual Bronchoscopy - Abstract:
- The current state-of-the-art workflow for endoscopic procedures consists of two steps. The first step involves acquisition and analysis of a three-dimensional computed tomography image to locate suspicious regions of interest and plan routes to reach them. In the second step, the physician attempts to maneuver the endoscope along the pre-planned route through the patient’s anatomy to reach these regions of interest. Navigation errors are common in standard practice and stem from the difficulty of pre-operative analysis and from the dissimilarity of the pre-operative data-set and the endoscopic video. This thesis proposes automated registration and tracking methods for registering the pre-operative data-set to the incoming live video in real-time. The registration locates the bronchoscope with respect to the anatomy and allows navigation routes and regions of interest that are defined pre-operatively within the computed tomography image to be continuously fused with the endoscopic video. The registration and tracking methods form the basis for an endoscopic guidance system described in this thesis. The guidance system uses novel interaction and visualization strategies that can assist a physician in reaching a region of interest and improve the accuracy of hitting a pre-specified anatomical target. We provide numerous laboratory results characterizing the performance of the registration methods. These results show that the registration algorithm is over 100x faster than previous methods, while simultaneously improving robustness and accuracy. We also present results showing the utility of the guidance system and strategies in phantom studies and in live human procedures.