3D-2D Operation Image Alignment with AI Projection Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing medical operation navigation systems face challenges in accurately aligning three-dimensional pre-operative images with two-dimensional intra-operative images, leading to alignment errors and prolonged registration times.
Innovation Solution
An operation image alignment method utilizing an artificial intelligence model trained with historical three-dimensional and two-dimensional image data sets to predict rotation and translation angles, employing image projection conversion technologies and model algorithms like generative adversarial networks and deep iterative 2D/3D registration, to align actual three-dimensional and two-dimensional images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image alignment methods are used to align three-dimensional and two-dimensional images, then the alignment process can be completed, but alignment errors occur and processing time is prolonged
Solution Approach 1:
The system performs preliminary actions by pre-processing the three-dimensional image to generate multiple two-dimensional projection images with different perspectives before the actual alignment operation. This preparation allows the system to have pre-computed reference data ready for rapid comparison and selection during the alignment process, reducing real-time processing requirements and improving both precision and speed.
Solution Approach 2:
The system creates multiple copies of the three-dimensional image in the form of two-dimensional projection images with different perspectives. These projection images serve as reference templates that can be quickly compared with the actual two-dimensional intra-operative image to determine the best match, thereby reducing alignment errors and processing time through pre-computed visual references.
2Measurement precision
If traditional image alignment methods are used, then the alignment process can be completed, but alignment errors occur
Solution Approach 1:
The system segments the three-dimensional image into multiple two-dimensional projection images with different perspectives (e.g., front view, side view, top view). This segmentation allows the system to compare specific visual features from different angles with the actual intra-operative image, improving alignment precision by evaluating multiple viewpoints simultaneously rather than relying on a single complex alignment algorithm.
Solution Approach 2:
The system introduces two-dimensional projection images as intermediary reference templates between the three-dimensional pre-operative image and the actual two-dimensional intra-operative image. These projection images serve as mediators that bridge the gap between different image modalities, enabling accurate alignment through visual matching without requiring complex direct transformation algorithms.
Data Source
AI summary
An operation image alignment method includes: inputting an actual three-dimensional image and an actual two-dimensional image of an actual desired part; converting the actual three-dimensional image into multiple sets of two-dimensional projection images according to image parameters of the actual three-dimensional image through an image alignment prediction model; comparing each of the sets of two-dimensional projection images to the actual two-dimensional image to calculate an image parameter difference value for each of the sets of two-dimensional projection image, and selecting one of the sets of two-dimensional projection image to obtain a predicted rotation angle and translation. The one of the sets of two-dimensional projection image has an image parameter difference value matching a preset difference value. The multiple sets of historical images containing at least one historical three-dimensional image and at least one historical two-dimensional image are used as a training data set of the image alignment prediction model.