Template-Based Anatomical Segmentation of Medical Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-atlas segmentation techniques for medical images face high computational costs and often result in less accurate atlas propagation when using a common template for all target images, leading to reduced segmentation accuracy and increased processing time.
Innovation Solution
A template library is employed to select optimal templates for each target image, reducing the number of online registrations required and utilizing offline registrations to achieve faster and more accurate anatomical segmentation by performing pairwise deformable registration with a subset of selected templates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multi-atlas segmentation is performed using standard techniques, then segmentation accuracy is improved, but computational cost and processing time increase significantly
Solution Approach 1:
The patent pre-computes and stores similarity metrics between each template and all target images before actual segmentation. This preliminary action allows the system to quickly retrieve and select appropriate templates during segmentation without performing computationally intensive similarity calculations at runtime, thereby maintaining accuracy while reducing processing time
Solution Approach 2:
The patent divides the large set of templates into multiple subsets or groups. Instead of comparing against all templates, the system selectively processes only relevant template subsets based on preliminary similarity metrics or anatomical characteristics, reducing the computational burden while preserving segmentation accuracy through targeted template selection
2Device complexity
If a common template is used for all target images, then device complexity is reduced, but segmentation accuracy deteriorates due to less accurate atlas propagation
Solution Approach 1:
The patent implements a dynamic template selection mechanism that adapts the template set based on each target image's characteristics. The system automatically identifies and selects the most appropriate templates for each specific target image rather than using a fixed common template, allowing the system complexity to increase only when and where it improves accuracy
Solution Approach 2:
The patent applies different templates or template subsets to different regions or types of target images based on their specific anatomical characteristics. This local customization ensures that each target image receives the most appropriate template matching its unique features, improving segmentation accuracy without requiring all possible templates to be processed for every image
3Measurement precision
If the number of templates is increased, then segmentation accuracy is improved, but computational cost increases
Solution Approach 1:
The patent performs partial template processing by selectively applying templates only to relevant regions or cases. The system identifies and processes only the necessary subset of templates for each target image rather than exhaustively processing all available templates, achieving sufficient segmentation accuracy with reduced computational effort
Solution Approach 2:
The system pre-evaluates and ranks templates based on their relevance to target images before actual segmentation. This preliminary filtering creates a prioritized template list that guides subsequent processing, ensuring that computational resources are focused on the most promising templates while less relevant ones are skipped or processed minimally
Data Source
AI summary
A mechanism is provided in a data processing system comprising a processor and a memory, the memory comprising instructions executed by the processor to specifically configure the processor to implement a multi-atlas segmentation engine. An offline registration component performs registration of a plurality of atlases with a set of image templates to thereby generate and store, in a first registration storage device, a plurality of offline registrations. The atlases are annotated training medical images and the image templates are non-annotated medical images. The multi-atlas segmentation engine receives a target image. An image selection component selects a subset of image templates in the set of image templates based on the target image. An online registration component performs registration of the subset of image templates with the target image to generate a plurality of online registrations. The multi-atlas segmentation engine retrieves offline registrations corresponding to the subset of image templates from the first registration storage device. The multi-atlas segmentation engine performs segmentation of the target image based on the retrieved offline registrations corresponding to the subset of image templates and the plurality of online registrations. The segmentation applies labels to anatomical structures present in the target image based on the retrieved offline registrations and the plurality of online registrations to thereby output a modified target image.