![Figure 1. Illustration of the 2D-3D registration.](/news/sites/default/files/styles/ucws_image_desktop/public/2020-08/Figure1_1.png?itok=SvmPgAZT)
Aug. 25, 2020
2D-3D Registration for a Dual Fluoroscopy (DF) Imaging System
![Geomatics Engineering Research](/news/sites/default/files/styles/ucws_news_hero_image_desktop/public/2020-08/bannerR_2.png?itok=C1ZMOLMB)
Dual fluoroscopy (DF) is an X-ray based non-invasive imaging system that can derive dynamic bony translations and rotations. The system consists of two high-speed video cameras, two image intensifiers, and two X-ray sources. By using low-dose X-rays, both high spatial and temporal resolution can be obtained, which yields accurate measurements. The 2D-3D registration is the process that matches 3D bone model acquired from MRI or CT scans with the 2D X-ray image pairs. Figure 1 shows the 2D-3D registration that estimates the pose of the 3D bone model from the image pair at each frame. This study is focused on the registration of MRI data as it can acquire detailed soft tissue contrast that is difficult to be seen from CT scans. With the registration of 3D bone model for both tibia (shank bone) and femur (thigh bone), tibiofemoral soft tissue deformation can be quantified for early detection of knee osteoarthritis (OA).
![Figure 1. Illustration of the 2D-3D registration.](/news/sites/default/files/styles/ucws_image_desktop/public/2020-08/Figure1_1.png?itok=SvmPgAZT)
The main objective of this research is to develop a 2D-3D registration system that can automatically reconstruct the MRI based 3D anatomical structure with submillimeter or higher accuracy. It includes four parts: 1) 2D-3D registration method development that is suitable for the MRI-based bone models with high precision and high efficiency, 2) Dynamic estimation to model how the bones translate and rotate over time to improve estimation accuracy, 3) Registration accuracy evaluation that requires the marker-based registration to provide the ground truth using implanted or imbedded beads, 4) Development of an automatic registration system that includes automatic initialization, registration, evaluation, and provides a visualization and user interaction tool for visual intervention when desired.
This work is supported by the Alberta Innovates Graduate Students scholarship fund, the T. Chen Fong Doctoral Research Excellence Scholarship fund in Medical Imaging Science, and the Natural Sciences and Engineering Research Council of Canada.