Breast Tissue Segmentation in Thermal Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automated segmentation of breast tissue from surrounding tissue in thermal images for breast cancer screening is challenging due to the amorphous nature of the tissue, making it difficult for software tools to accurately identify regions of interest, especially when screening a large number of patients.
Innovation Solution
A method involving a thermal image analysis to identify a set of points based on the contour of the body and isotherms, connecting these points to form a polygon that segments the breast from surrounding tissue, with draggable vertices for user adjustment, and communication of the segmented area to a software tool for further analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If expert visual inspection is used to segment breast tissue, then segmentation accuracy is improved, but productivity deteriorates due to inability to handle large patient volumes
Solution Approach 1:
The system enables semi-automated segmentation where the software automatically identifies and segments breast tissue regions, reducing reliance on expert manual inspection while maintaining accuracy. The algorithm processes thermal images autonomously to generate segmented regions of interest.
Solution Approach 2:
The patent replaces the mechanical visual inspection process with an automated image processing system that uses thermal image analysis, isotherm detection, and polygon-based segmentation algorithms to automatically identify and segment breast tissue without requiring manual expert intervention.
2Productivity
If automated segmentation software is implemented, then productivity is improved, but measurement precision deteriorates due to difficulty in accurately identifying regions of interest
Solution Approach 1:
The system introduces draggable control points as an intermediary between automated segmentation and final region definition. These control points allow users to adjust and refine the automatically generated polygon boundaries, ensuring accurate region of interest identification while maintaining high productivity.
Solution Approach 2:
The segmentation system transitions from static automated polygon generation to dynamic adjustable boundaries. Users can interactively modify control points and polygon shapes to adapt to specific anatomical variations, improving measurement precision without sacrificing automated processing efficiency.
3Measurement precision
If manual segmentation by experts is used, then measurement precision is improved, but loss of time increases when screening many patients
Solution Approach 1:
The system performs preliminary automated segmentation to generate initial region of interest polygons before user review. This pre-processing step creates a starting point that is already close to the final accurate segmentation, reducing the time experts need to spend on manual adjustment while maintaining high measurement precision.
4Productivity
If sophisticated automated software tools are developed, then productivity is improved, but device complexity increases
Solution Approach 1:
The software is divided into distinct functional modules: thermal image acquisition, isotherm detection, control point identification, polygon generation, and user interaction handling. This modular segmentation allows each component to be independently optimized and maintained, managing overall system complexity while enabling sophisticated automated segmentation capabilities.
Data Source
AI summary
What is disclosed is a system and method for automatically segmenting a breast from surrounding tissue in a thermal image. A thermal image of at least one breast of the patient is received. The thermal image is then analyzed to identify a set of N points around the breast, a contour of an outline of the body, and isotherms of the axilla and infra-mammary fold. Thereafter, the points are connected together to form a N-sided irregular polygon which segments the breast from surrounding tissue. Each of the points is a vertex of the polygon and comprises a draggable object which enables a user to selectively manipulate a shape of the polygon. A user can add/delete vertices from the polygon as desired. The area of the image encompassed by the polygon is communicated to a breast cancer screening algorithm performing automated or semi-automated screening.