A novel machine learning model for breast cancer detection using mammogram images

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Breast cancer is a leading cause of death in women globally.
• Early detection significantly improves recovery rates and reduces mortality.
• Deep learning shows promise for advancing breast cancer screening technologies.

Purpose of the Study:

• To propose a novel machine learning method for breast cancer detection in mammograms.
• To enhance feature extraction, classification, and optimization for improved diagnostic accuracy.

Main Methods:

• Image preprocessing including noise removal, smoothing, and normalization.
• Feature extraction using probabilistic principal component analysis.
• Classification via Naïve Bayes and transfer integrated convolution neural networks (TCNN).
• Optimization using firefly binary grey optimization (FBGO) and metaheuristic moth flame lion optimization (MMFLO).
• Ensemble model combining classifiers and optimizers.

Main Results:

• The proposed Bayes+FBGO classifier achieved 95% accuracy on the INbreast dataset.
• The TCNN+MMFLO classifier achieved 98% accuracy on the INbreast dataset.
• The ensemble model demonstrated robust performance across diverse mammography datasets.

Conclusions:

• The developed machine learning framework offers a highly accurate approach for breast cancer detection.
• The combination of advanced classifiers and optimizers shows significant potential for improving mammogram analysis.
• This research contributes to the advancement of AI-driven tools for early breast cancer diagnosis.