Adaptive Thresholding for FG Tissue Segmentation in Mammography
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for fibro-glandular tissue segmentation in digital mammography face challenges due to varying breast densities and similarities in brightness patterns between FG tissues and other anatomic structures, leading to inaccurate segmentation, especially in low and high density breasts.
Innovation Solution
An automated method using a fuzzy logic framework that extracts global and specific features from mammography images, applies an unsharp mask and normalization, and configures a fuzzy logic module to compute an adaptive threshold for accurate FG tissue segmentation, accounting for varying breast densities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a constant threshold method is used for FG tissue segmentation, then the device complexity is low, but the segmentation accuracy deteriorates (67.4% accuracy)
Solution Approach 1:
The patent implements an adaptive thresholding mechanism that dynamically adjusts segmentation thresholds based on local image characteristics and breast density estimates. Instead of using a fixed constant threshold, the system calculates density-specific thresholds that adapt to varying tissue densities across different regions of the mammogram, thereby improving segmentation accuracy while managing complexity through automated adaptation.
Solution Approach 2:
The patent changes the threshold parameter from a constant value to a density-dependent variable. By estimating breast density first and then selecting thresholds based on the estimated density category (fatty, glandular, or dense), the system optimizes segmentation accuracy for different tissue types. This parameter change allows the same segmentation algorithm to perform well across diverse breast densities.
2Measurement precision
If breast density variation is considered in segmentation, then segmentation accuracy improves, but the difficulty of detecting and measuring increases
Solution Approach 1:
The patent performs preliminary breast density estimation before conducting FG tissue segmentation. By first categorizing the breast density (fatty, glandular, or dense) using global image features and histograms, the system prepares density-appropriate parameters and thresholds in advance. This preliminary action simplifies the subsequent segmentation step by providing density-specific guidance, thereby reducing the overall difficulty of detection despite considering density variations.
Solution Approach 2:
The patent introduces breast density estimation as an intermediary step between image acquisition and FG tissue segmentation. This intermediary process uses global image characteristics to determine tissue density, which then guides the segmentation process by providing density-specific thresholds and parameters. This mediator simplifies the complex task of segmenting across all densities by breaking it into manageable density-specific subtasks.
3Reliability
If fuzzy logic module is configured with manual settings, then the reliability of segmentation improves, but the ease of operation deteriorates
Solution Approach 1:
The patent implements automated configuration of the fuzzy logic module that performs self-calibration using training data. The system automatically learns optimal membership functions and rule parameters from labeled mammogram images, eliminating the need for manual expert configuration. This self-service approach maintains high reliability through data-driven optimization while significantly improving ease of operation by removing manual setup requirements.
4Measurement precision
If feature extraction and fuzzy logic processing are applied, then segmentation accuracy improves, but processing time increases
Solution Approach 1:
The patent divides the processing pipeline into distinct modular stages: breast density estimation, feature extraction, fuzzy logic inference, and final segmentation. Each stage processes specific features and produces intermediate results that feed into the next stage. This segmentation of processing allows for optimized computation at each step and enables parallel processing where applicable, reducing overall processing time while maintaining accuracy.
Data Source
AI summary
Embodiments of the present invention provide automated systems and methods for segmentation of fibro-glandular (FG) tissue in digital mammography. A classifier is trained for breast density without the prior knowledge for FG tissue. The classifier is then used for feature selection, where the selected features are fed into a fuzzy logic module, and an adaptive threshold is obtained. Post-processing is performed on the image in order to reduce regions which may have been misclassified during the FG segmentation.