Breast Cancer Detection Using Cascade Classifiers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computerized systems for breast cancer detection in x-ray mammography face high false-positive and false-negative rates, requiring extensive radiologist time for examination and diagnosis.
Innovation Solution
A cascade of multiple classifiers is employed, utilizing a deep-learning network trained on non-x-ray data to extract features and classify patches, with regression and patch classification methods to reduce false positives and negatives, and a machine-learned classifier to provide accurate breast cancer indications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If computerized systems use hand-crafted features or traditional image processing methods, then the system can be implemented with current technology, but the false-positive rate becomes high making it impractical
Solution Approach 1:
The patent replaces traditional hand-crafted feature extraction and classification mechanisms with deep learning-based automated systems. The deep learning model automatically learns optimal features from image data, eliminating the need for manual feature engineering and reducing false positives through superior pattern recognition capabilities.
Solution Approach 2:
The patent transforms the detection approach by changing from traditional image processing parameters to deep learning model parameters. The system uses neural network layers that process image data through multiple transformations, enabling more accurate cancer detection while reducing false positives through sophisticated feature learning.
2Reliability
If radiologists examine all mammograms manually, then detection accuracy can be maintained, but the workload increases requiring greater efficiency from radiologists
Solution Approach 1:
The patent segments the mammogram examination process into two parts: automated deep learning classification that handles initial screening and triage, and radiologist review focused only on suspicious areas. This segmentation allows radiologists to maintain high accuracy while improving efficiency by not examining every pixel manually.
Solution Approach 2:
The deep learning system performs self-service by automatically screening entire mammograms and prioritizing cases requiring radiologist attention. The system serves itself by learning from training data and making independent classification decisions, reducing the burden on radiologists while maintaining detection accuracy.
3Device complexity
If a single classifier is used for breast cancer detection, then the system is simple, but false-positive and false-negative rates remain high
Solution Approach 1:
The patent divides the classification task into multiple stages using a cascade of classifiers. The first classifier performs initial screening, and subsequent classifiers refine the results. This segmentation of the classification process improves detection accuracy while keeping each individual classifier relatively simple.
Solution Approach 2:
The patent applies preliminary classification actions using a first classifier that screens all mammograms before they reach the second classifier. This preliminary action filters out obvious negative cases and prioritizes suspicious areas, improving overall system accuracy without requiring every classifier to be overly complex.
Data Source
AI summary
For breast cancer detection with an x-ray scanner, a cascade of multiple classifiers is trained or used. One or more of the classifiers uses a deep-learnt network trained on non-x-ray data, at least initially, to extract features. Alternatively or additionally, one or more of the classifiers is trained using classification of patches rather than pixels and/or classification with regression to create additional cancer-positive partial samples.