Deep learning techniques for breast mass malignancy classification on digital mammography
Ariel Coto Santiesteban, Lisbel Garzón Cutiño, Damián Valdés Santiago · Salud Ciencia y Tecnología - Serie de Conferencias · 2024
Introduction: Breast cancer is one of the most common type of cancer with a high mortality rate. Mammography is widely used to identify breast cancer. Computer Aided Diagnosis systems are used for automatic detection of breast lesions. Methods: We propose and evaluate a deep learning model, called VGG16-C300, for breast mass malignancy classification. CBIS-DDSM dataset was used for training and evaluation. Image contrast enhancement methods like CLAHE and Mean Blur where previously applied to regions of interests. Results: The trained model achieved and area under the curve of 0.80, after 10 iterations of a 5-fold Cross-Validation. Conclusions: VGG16-C300 could be used as a component in a computer-aided diagnosis system for breast cancer detection