Multi-atlas Segmentation Using Landmark-Based Subset Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multi-atlas segmentation methods for medical images require extensive processing time and computational resources, especially when a large number of anatomical variations are involved, making them impractical for real-time applications.
Innovation Solution
A medical image processing apparatus that selects a subset of atlases based on anatomical landmarks using a distance metric and clustering methods, performing a less computationally intensive initial registration followed by a more accurate non-rigid registration for improved segmentation efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large number of atlases are used to improve segmentation accuracy by covering more anatomical variations, then segmentation accuracy is improved, but processing time and computational resources become prohibitively long and intensive
Solution Approach 1:
The patent segments the large set of atlases into subsets based on anatomical similarity. Instead of processing all atlases uniformly, the system divides them into groups and selects representative subsets, reducing the computational burden while maintaining segmentation accuracy across different anatomical variations.
Solution Approach 2:
The patent applies partial action by selecting only a subset of atlases that are most relevant to the specific image being segmented. Rather than using all available atlases (excessive action), the system identifies and uses only the necessary portion that provides sufficient anatomical coverage for accurate segmentation.
2Manufacturing precision
If multiple registration methods are performed on all atlases to improve segmentation quality, then segmentation quality is improved, but computational complexity and processing time increase significantly
Solution Approach 1:
The patent applies local quality by performing different registration methods on different subsets of atlases rather than applying the same comprehensive registration to all atlases. The system performs initial registration on all atlases, then applies more computationally intensive refined registration only to selected subsets, optimizing computational resources while maintaining segmentation quality.
Solution Approach 2:
The patent performs preliminary registration of all atlases to a reference atlas before selecting subsets for refined registration. This preliminary action establishes initial alignment and enables subsequent selective processing, reducing overall computational complexity while maintaining segmentation quality.
3Measurement precision
If comprehensive registration of all atlases is performed to ensure accurate anatomical alignment, then anatomical alignment accuracy is improved, but processing efficiency deteriorates
Solution Approach 1:
The patent segments the atlas population into subsets based on anatomical similarity metrics. By dividing the comprehensive registration task into smaller subset-specific registrations, the system maintains anatomical alignment accuracy for each subset while improving overall processing efficiency through parallel and selective processing.
Solution Approach 2:
The patent changes the parameter of atlas selection by using anatomical similarity metrics to identify and select subsets of atlases that are most relevant to the target image. This parameter change enables the system to maintain high anatomical alignment accuracy by selecting appropriate atlases while improving processing efficiency by excluding irrelevant ones.
Data Source
AI summary
An image data processing apparatus including a data receiver receiving image data to be segmented, and an atlas selection processor accessing a plurality of atlas data sets and selecting a subset of the atlas data sets for use in segmenting the image data, wherein the atlas selection processor is configured to select the subset of atlas data sets in dependence on the positions of one or more anatomical landmarks comprised in the plurality of atlas data sets.