Image Segmentation Using ROI Templates and Automated Model Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image segmentation methods are inefficient and prone to human error due to the need for manual retrieval and training of segmentation models, requiring extensive manpower and time for high-quality labelled samples.
Innovation Solution
A method and system for image segmentation that determines a segmentation range and template from a target image, using a list of regions of interest (ROIs) to segment target portions with improved efficiency and accuracy, and automatically labels un-labelled samples to generate high-quality labelled samples for training segmentation models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual retrieval and training of segmentation models is performed, then segmentation accuracy can be maintained, but efficiency deteriorates and human error increases
Solution Approach 1:
The system automatically determines segmentation ranges and selects appropriate segmentation models without requiring manual user intervention. The automated segmentation model selection module retrieves pre-trained segmentation models based on the determined segmentation range, eliminating the need for manual model retrieval and training while maintaining segmentation accuracy.
Solution Approach 2:
The system pre-determines segmentation ranges and pre-selects segmentation models before actual segmentation processing. By determining the segmentation range and retrieving the corresponding segmentation model in advance, the system eliminates time-consuming manual retrieval and training processes, thereby improving efficiency while maintaining accuracy.
2Measurement precision
If manual labelling of training samples is performed, then high-quality labelled samples can be obtained, but manpower and time requirements increase
Solution Approach 1:
The system automatically labels unlabelled images by determining segmentation ranges and selecting segmentation models without requiring manual labelling. The automated process generates high-quality labelled samples by using pre-trained segmentation models to process images, thereby reducing manpower and time requirements while maintaining sample quality.
Solution Approach 2:
The system creates labelled samples by automatically applying pre-trained segmentation models to unlabelled images, effectively copying the labelling process performed by experts. This automated copying approach generates high-quality labelled samples without requiring manual intervention, significantly reducing time and manpower costs.
3Measurement precision
If multiple segmentation models are trained for different ROIs, then segmentation accuracy for specific regions improves, but system complexity increases
Solution Approach 1:
The system uses a unified automated segmentation model selection module that can handle multiple types of ROIs (tumors, organs at risk, etc.) through a single interface. The module automatically determines segmentation ranges and selects appropriate pre-trained models, providing universal functionality for different ROI types without requiring separate manual processing workflows, thereby reducing system complexity.
Solution Approach 2:
The system segments the segmentation process into distinct automated stages: determining segmentation range, selecting segmentation model, and performing segmentation. By dividing the complex model management task into separate automated modules, the system maintains the ability to handle multiple ROI types with high accuracy while reducing overall system complexity through modular automation.
Data Source
AI summary
The present disclosure provides systems and methods for image segmentation. The methods may include determining, from a target image of a subject, a segmentation range. The methods may include determining a segmentation template corresponding to the segmentation range. The segmentation template may include a list of one or more regions of interest (ROIs) of the subject in the segmentation range. Further, the methods may include segmenting one or more target portions corresponding to the one or more ROIs from the target image using at least one segmentation model corresponding to the segmentation template.