3D Medical Image Alignment Using Common-Structure Centroids
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for aligning three-dimensional medical images, such as CT and MRI images, are computationally intensive and may not accurately align images acquired by different imaging apparatuses, leading to inefficiencies and increased burden on radiologists.
Innovation Solution
An image alignment apparatus that derives three-dimensional coordinate information for common structures in multiple images and aligns them by setting bounding boxes around these structures, using centroid positions to facilitate quick and accurate alignment, even when images are from different imaging devices or acquired at different times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If rigid body alignment or non-rigid body alignment method is used to align three-dimensional images, then alignment accuracy can be improved, but the amount of calculation increases and processing time becomes longer
Solution Approach 1:
The patent segments the alignment process into two distinct stages: a rough alignment stage using simple feature point matching, and a fine alignment stage using the segmented structure information. This segmentation allows the system to achieve high alignment accuracy through the fine alignment stage while keeping the overall processing time short by using the computationally efficient rough alignment stage for initial positioning.
Solution Approach 2:
The patent performs preliminary rough alignment using feature point matching before conducting fine alignment using segmented structure information. This preliminary action establishes an initial alignment state that reduces the search space for the subsequent fine alignment process, thereby achieving high accuracy without requiring extensive computation time.
2Measurement precision
If rigid body alignment or non-rigid body alignment method is used to align three-dimensional images, then alignment accuracy can be improved, but the amount of calculation increases
Solution Approach 1:
The patent segments the alignment task into rough alignment and fine alignment components. The fine alignment uses segmented structure information (such as organ boundaries or anatomical landmarks) to achieve high precision, while the rough alignment uses simpler feature point correspondence. This segmentation reduces calculation complexity by avoiding the need to perform complex optimization throughout the entire alignment process.
Solution Approach 2:
The patent performs preliminary rough alignment to establish an initial transformation matrix before conducting fine alignment. This preliminary action simplifies the subsequent fine alignment calculation by reducing the search space and initializing the optimization process closer to the optimal solution, thereby reducing overall calculation complexity.
3Adaptability or versatility
If existing alignment methods are used for images from different imaging apparatuses, then alignment can be performed, but the alignment accuracy decreases
Solution Approach 1:
The patent introduces segmented structure information as an intermediary element that bridges images from different imaging apparatuses. By using anatomically meaningful structures (such as organ contours or landmark points) as the basis for fine alignment, the system can accurately register images from different modalities (e.g., CT and MRI) that have different imaging characteristics, thereby maintaining high alignment accuracy across diverse imaging devices.
Data Source
AI summary
An image alignment apparatus includes at least one processor, and the processor derives, for each of first and second three-dimensional images each including a plurality of tomographic images and a common structure, first and second three-dimensional coordinate information that define an end part of the structure in a direction intersecting the tomographic image. The processor aligns the first three-dimensional image and the second three-dimensional image by using the first and second three-dimensional coordinate information to align the common structure included in each of the first three-dimensional image and the second three-dimensional image at least in the direction intersecting the tomographic image.