5 Years Impact Factor: 1.53
Author: Gujjari Usharani, Sharwani Nune, Taili Saritha, Gayam Swathi
Abstract:
Early identification of breast cancer relies on mammogram detection, which may be greatly improved with the use of machine learning algorithms. Classification and segmentation of mammography pictures for abnormality detection is the focus of this study. In order to mark mammography pictures as "Normal" or "Malignant," we used a Convolutional Neural Network (CNN) technique for binary classification. We used an Attention-based optimised UNET algorithm for region-based abnormality detection segmentation. The dataset was preprocessed for training by shrinking, normalising, and shuffling. It is impossible to download or handle entire mask image datasets (166 GB) using normal methods, thus we trained with a reduced selection of photos. As a consequence, the segmentation results may not be as precise as expected because of the insufficient training data, even while the classification model performs reasonably. When applied to test photos, the trained model can detect cancerous area
Download PDF