Enhanced Comparative Study of Deep Learning and Hybrid Models for Breast Cancer Histopathology Image Classification

Abstract

Using mammography images from the CBIS-DDSM (Curated Breast Imaging Subset of the Digital Database for Screening Mammography), an extensive dataset of digitized mammograms, deep learning techniques are used in this work to classify breast cancer. A total of 277,524 50 x 50-pixel image patches were extracted and labeled according to whether they were malignant. This dataset was used to train a custom Convolutional Neural Network (CNN) that can classify regions as either cancerous or non-cancerous. With a 96.81% accuracy rate on the test set, the model showed excellent performance and generalizability. With potential uses in computer-aided diagnosis systems, this study highlights the efficacy of CNN-based models in automated breast cancer detection from mammography.