A CNN-Based System for Breast Cancer Detection in Mammograms

A CNN-Based System for Breast Cancer Detection in Mammograms

Dr. Aditi V. Vedalankar1, Saroj Mali2, Vaishnavi Patel3, Arpita Sathe4

Dept of Electronics & Telecommunication Engineering JSPM’s RSCOE Tathawade Pune.

avvedalankar@gmail.com sarojmali85@gmail.com patelvaishnavi43@gmail.com arpitasathe2002@gmail.com

Abstract—Cancer of the breast is the second most common cancer globally, and it is far more common among women. Continuous advancements in medical imaging are crucial for early detection, leading to improved treatment outcomes and reduced mortality rates. Machine learning offers a promising avenue for enhancing diagnostic accuracy, with deep learning, specifically neutral networks, proving to be a valuable technique for classifying benign or malignant breast conditions. Known as convolutional neural networks, or CNNs, employed in this study using a dataset of mammograms that encompasses both benign and malignant breast images. The experimental results that were obtained promote the integration of present- day extraction and classification of features methods in a variety of medical imaging uses especially those connected with breast cancer, by revealing the potency of deep learning methods for identifying malignancy from mammogram images. To further improve accuracy measurements, research is currently being done on already trained networks and CNN built modification. For simplified extraction and sorting of features services, precise segmentation is still critical.

Keywords—Mammogram, Breast Cancer, Curated Breast Imaging Subset (CBIS), Digital Database for Screening Mammography (DDSM), Convolution Neural Network (CNN)