Evaluation of a New Ensemble Learning Framework for Mass Classification in Mammograms

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Biomedical Engineering

Background:

• Mammography is a primary screening tool for breast cancer diagnosis.
• Accurate differentiation between benign and malignant masses is crucial.

Purpose of the Study:

• To develop and evaluate a computer-aided diagnosis (CAD) system for breast mass classification in mammograms.
• To enhance diagnostic accuracy and efficiency in breast cancer screening.

Main Methods:

• Implementation of a CAD system involving noise removal, mass segmentation using deformable models, and feature extraction (shape, border, tissue properties, fractal dimension).
• Utilized a genetic algorithm for optimal feature subset selection.
• Developed a novel classifier combination architecture that differentiates and trains easy and difficult samples separately.

Main Results:

• The proposed CAD system demonstrated competitive accuracy compared to existing state-of-the-art methods.
• The system was validated on the mini-Mammographic Image Analysis Society (MIAS) and Digital Database for Screening Mammography (DDSM) databases.

Conclusions:

• The developed CAD system shows significant potential for improving breast cancer diagnosis accuracy.
• The novel approach to feature selection and classifier combination contributes to advancing computer-aided detection in mammography.