Breast Image CAD Risk Scoring for Hard-to-Detect Cancer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current breast cancer risk prediction models do not adequately incorporate high-resolution breast image information, such as microcalcifications and masses, leading to suboptimal risk assessment and missed detections.
Innovation Solution
A deep learning-based CAD system analyzes breast imagery to detect microcalcifications and other lesions, combining these features with genetic, lifestyle, and familial risk factors to generate accurate risk scores, including a masking score to indicate undetectable tumors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional risk prediction models are used without breast imagery analysis, then the system complexity is low, but the risk prediction accuracy is insufficient
Solution Approach 1:
The patent combines traditional risk prediction models with CAD-based breast imagery analysis into a unified risk assessment system. The CAD system detects microcalcifications and masses, and these detection results are integrated with genetic, lifestyle, and familial risk factors to generate comprehensive risk scores, thereby improving prediction accuracy while managing system complexity through modular integration.
2Measurement precision
If high-resolution breast image information is incorporated into risk models, then the risk assessment accuracy improves, but the detection and measurement difficulty increases
Solution Approach 1:
The patent replaces manual detection and measurement of breast image features with an automated CAD (computer-aided detection) system. The CAD system uses image processing algorithms to automatically detect microcalcifications, masses, and other lesions, eliminating the need for manual analysis and significantly reducing the difficulty of detecting and measuring subtle breast abnormalities while improving risk assessment accuracy.
3Reliability
If only traditional risk factors are used without CAD analysis, then the system is easier to operate, but microcalcifications and masses may be missed
Solution Approach 1:
The patent implements a self-service CAD system that automatically performs detection of microcalcifications and masses without requiring manual intervention. The system autonomously analyzes breast images, generates detection results, and integrates them with risk factors to produce risk assessments, thereby maintaining high detection reliability while preserving ease of operation through automation.
4Measurement precision
If breast imagery analysis is added to risk models, then the AUC improves by 7%, but the processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary detection of microcalcifications and masses using the CAD system before conducting the full risk assessment. By pre-identifying suspicious lesions and extracting relevant features in advance, the system reduces the computational burden during the final risk calculation phase, thereby improving AUC while minimizing the loss of time through efficient preprocessing.
Data Source
AI summary
This invention provides a system and method for assessing risk of a breast cancer diagnosis based upon imagery of tissue and (optionally) other patient-related factors. A CAD (or similar) system analyzes the imagery and generates a plurality of numerical feature values. An assessment module receives inputs from patient factors and history and computes the risk based upon the feature values and the patient factors and history. A masking module receives inputs from the patient factors and history, and computes the risk of having a cancer, which cancer is otherwise characterized by a low probability of detection, based upon the feature values and the patient factors and history. A recall module receives inputs from the assessment module and the masking assessment module, and generates a computer-aided indication of a clinical follow-up by the patient. Results of assessment(s) can be displayed to the clinician and/or patient using a graphical interface display.