Novel ANN Based Algorithm for Accurate Detection of Breast Cancer

Abstract

Breast cancer is a disease influenced by genetic and environmental factors, exhibiting uncontrolled proliferation of abnormal cells within the breast tissue. Early detection of breast cancer rely heavily on imaging modalities such as mammography, MRI, ultrasound, and fine-needle biopsy. It is one of the most prevalent cancers worldwide, with varying incidence rates across demographics and regions. Despite advancements in screening and treatment, challenges persist, emphasizing the need for ongoing research to refine diagnostic tools, develop individualized therapies, and ultimately reduce mortality associated with this disease. newlineGiven the exponential growth of medical data, developing effective and precise cancer classification systems is crucial for early detection and efficient treatment. This thesis explores how Convolutional Neural Networks (CNNs) and transfer learning techniques can improve the efficiency and accuracy of cancer classification from medical imaging data. newlineThe primary goal of this research is to investigate the performance of CNN architectures with transfer learning for classifying various cancer types, including prostate, lung, and breast cancer. Transfer learning allows us to adapt knowledge gained from large-scale image datasets like ImageNet to the specific task of cancer classification. newlineThe methodology involves collecting and preprocessing diverse histopathological images. Several CNN architectures serve as base networks for transfer learning, including state-of-the-art models such as ResNet, ImageNet, and InceptionNet. Comprehensive experiments evaluate these models in terms of accuracy, sensitivity, specificity, and computational efficiency. newlineThe thesis also examines the impact of hyperparameter optimization and data augmentation techniques on classification performance. To reduce overfitting and enhance generalization, methods like batch normalization and dropout regularization are investigated. newlineFindings demonstrate that transfer learning significantly boosts cancer classification accuracy