2D-3D Surgical Registration for Joint Posture Changes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 2D-3D registration techniques in computer-assisted surgery require significant human-computer interaction and are prone to errors due to changes in joint posture during surgery, leading to reduced accuracy and precision.
Innovation Solution
A method and system that perform posture transformation on a 3D image based on 2D images captured during surgery to achieve rough and fine registrations, optimizing transformation matrices without manual selection of anatomical points, using modules for image acquisition, posture transformation, and target matrix determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual selection of anatomical points and regions is performed for 2D-3D registration, then the registration process can be completed, but the accuracy and efficiency are reduced due to complex human-computer interaction operations
Solution Approach 1:
The system automatically identifies anatomical landmarks and performs registration without requiring manual selection by the operator. The computer vision algorithms detect and match anatomical points between 2D and 3D images autonomously, eliminating complex human-computer interaction while maintaining or improving registration accuracy
Solution Approach 2:
The manual mechanical interaction of selecting points and adjusting parameters is replaced with automated computer vision and image processing algorithms. The system uses computational methods to automatically establish correspondences between 2D and 3D anatomical structures, substituting human操作 with automated processing
2Measurement precision
If 2D-3D registration is performed on joint parts composed of multiple bones, then the surgical plan can be mapped to surgical space, but the precision is reduced due to relative motions of bones caused by joint posture changes
Solution Approach 1:
The system accounts for the dynamic nature of joint structures by capturing multiple 2D images at different joint postures and performing registration for each posture. The transformation matrix is optimized based on the specific joint configuration, allowing accurate mapping even when bones move relative to each other during surgery
Solution Approach 2:
The system adjusts registration parameters based on detected joint posture. By identifying the current joint configuration and selecting appropriate transformation parameters for that specific posture, the system maintains high precision despite variations in bone positions caused by joint movement
3Reliability
If multiple 2D images are captured and processed for registration, then the mapping relationship between 3D image space and surgical space can be established, but the time consumption increases due to frequent manual operations
Solution Approach 1:
The system processes multiple 2D images continuously through automated algorithms without requiring repeated manual intervention. Once the registration framework is established, subsequent images are processed automatically, maintaining continuous workflow and reducing the time loss associated with frequent human operations
Solution Approach 2:
The system performs preliminary automated processing of 2D images including feature detection, landmark identification, and initial alignment before final registration. This preliminary automated action reduces the need for time-consuming manual adjustments during the critical registration phase
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A method for registration, implemented by at least one processor and comprising: obtaining a 3D image of a target object captured before a surgery and at least one 2D image captured during the surgery (310); obtaining a registered 3D image and a first transformation matrix between a 3D image coordinate system and a surgical space coordinate system by performing posture transformation on the 3D image based on the at least one 2D image (320); determining a 2D target image corresponding to a target part of the target object in each of the at least one 2D image and determining a 3D target image corresponding to the target part in the registered 3D image (330); and obtaining a target transformation matrix by performing posture transformation on the registered 3D image based on the 3D target image and the 2D target image in each of the at least one 2D image to optimize the first transformation matrix (340).