Automated Mammographic Density Estimation Using Tissue Probability Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating mammographic density in mammograms are subjective and prone to high inter-observer and intra-observer variability due to qualitative estimation techniques, which affect the accuracy and reproducibility of breast cancer risk assessment.
Innovation Solution
An automated mammographic density estimation method using prior probability information, where a population-based tissue probability map is constructed by extracting tissue probability information from expert-segmented images, allowing for pixel-wise tissue probability calculation and accurate density estimation, incorporating learned knowledge from expert readers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If qualitative estimation method is used to classify mammographic density, then the method is simple and easy to operate, but the reliability and accuracy are reduced due to significant variation depending on observer's experience and physiologic condition
Solution Approach 1:
The patent replaces the manual qualitative estimation process (mechanical human observation and classification) with an automated computer-based system that performs pixel-wise tissue probability calculation and density estimation, thereby eliminating observer variability while maintaining operational simplicity
Solution Approach 2:
The system incorporates learned knowledge of expert readers into prior probability information that is automatically applied during density estimation, allowing the system to self-correct and improve accuracy without requiring continuous expert intervention
2Measurement precision
If automated computer-aided diagnosis method is used for objective and quantitative estimation, then the reliability and measurement precision are improved, but the device complexity increases
Solution Approach 1:
The patent performs preliminary segmentation of the breast area from the mammogram and constructs population-based tissue probability maps before the actual density estimation, which simplifies the main estimation process by pre-processing and organizing the data in advance
Solution Approach 2:
The system transforms the complex qualitative assessment into quantitative pixel-wise tissue probability calculations by changing the parameter representation from categorical classifications to continuous probability values, enabling more precise measurement while managing complexity through mathematical transformation
Data Source
AI summary
Disclosed is an automated mammographic density estimation method using statistical image information, the method including a preprocessing step of reading the mammogram, segmenting a breast area and shifting pixel values; a step of constructing a tissue probability map in which population-based probability information is extracted, and a probability map for glandular and adipose tissues is constructed; and a density estimation step in which a breast area is segmented based on the constructed tissue probability map and a mammographic density is calculated.