AI-Based 3D Humerus Reconstruction from 2D X-Ray Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer-assisted surgery techniques lack effective methods for accurately determining the entry point and correct positioning of intramedullary nails in long bones during minimally invasive procedures, particularly due to challenges in interpreting X-ray images and determining anatomical angles like anteversion and torsion, leading to potential malrotation and discomfort for patients.
Innovation Solution
A system and method that processes intraoperative 2D X-ray images to provide 3D geometric information without the need for reference bodies or trackers, using image registration and 3D reconstruction to determine the relative position and orientation of objects, including anatomical structures and implants, by combining features from multiple X-ray images taken from different angles and restricting allowable C-arm movements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional palpation and rule of thumb methods are used to determine entry point, then the procedure is simple and quick, but the accuracy and precision of entry point determination deteriorates substantially
Solution Approach 1:
The patent replaces the mechanical palpation method with an automated image processing system that uses 2D X-ray images and 3D bone models to determine the entry point. The system substitutes manual finger-based measurement with computer-aided detection algorithms that process radiographic images to identify the optimal entry point with high precision.
Solution Approach 2:
The patent creates a 3D digital copy or model of the patient's bone anatomy based on 2D X-ray images. This virtual 3D model serves as a replica that allows precise measurement and determination of the entry point without requiring complex physical manipulation or multiple invasive imaging procedures.
2Measurement precision
If multiple X-ray images from different directions are acquired to improve entry point determination, then the accuracy improves, but the radiation exposure and procedure time increase
Solution Approach 1:
The patent uses a minimal set of 2D X-ray images (often just one or two standard views) to reconstruct a 3D bone model and determine the entry point. Rather than requiring multiple images from every possible angle, the system extracts sufficient geometric information from limited projections to achieve accurate 3D localization, thereby reducing radiation exposure and procedure time.
Solution Approach 2:
The system performs preliminary 3D reconstruction of the bone model from 2D X-ray images before the actual implantation procedure. This pre-processing step creates a virtual anatomical model that guides the surgeon in determining the entry point, eliminating the need for time-consuming intraoperative measurements and adjustments.
3Measurement precision
If 3D reconstruction from 2D X-ray images is performed, then the accuracy of anatomical structure representation improves, but the computational complexity and processing requirements increase
Solution Approach 1:
The patent replaces complex manual measurement and visualization methods with automated computer-based 3D reconstruction algorithms. The system uses software to automatically process 2D X-ray images, perform geometric calculations, and generate 3D bone models, thereby reducing the need for complex manual operations while achieving high measurement precision.
4Measurement precision
If reference bodies or trackers are used for image registration, then the accuracy of 3D localization improves, but the invasiveness and complexity of the procedure increases
Solution Approach 1:
The patent extracts geometric information directly from the patient's existing 2D X-ray images without introducing external reference bodies or trackers into the surgical field. The system removes the need for additional invasive components by utilizing the anatomical structures themselves and the available radiographic data for 3D localization and image registration.
Data Source
AI summary
A system with a processing unit for processing X-ray images may be utilized for assisting in humerus surgery based on X-ray images. Firstly, at least two X-ray images are received having been generated with different imaging directions and showing a proximal portion of a humerus. Then, (i) the X-ray images are registered, (ii) an approximation of at least a part of the 2D outline of the humerus head in both images is determined, (iii) a 3D approximation of the humerus head based on the approximated 2D outlines and the registered images is determined, and (iv) 2D image coordinates of a total of at least three different points in the X-ray images are determined. Finally, an approximation of an anatomical neck is determined as a curve on the 3D approximation of the humerus head based on the at least three determined points.