IoT with Cloud Based Breast Cancer Diagnosis Using Deep Learning Techniques
G. Ranjith Kumar, M Ranjani, R. Santhiya, S.S. Thamilselvi · 2023
Breast cancer is a significant public health issue affecting women worldwide. Early detection and accurate diagnosis are crucial for improving survival rates and reducing mortality. Deep learning, an instance of machine learning, has yielded encouraging results in a variety of medical applications, especially breast cancer detection. Deep learning algorithms can automatically learn and identify patterns in medical images, such as mammograms, to detect breast cancer. Moreover, deep learning can also assist in predicting breast cancer sub-types and prognoses. This proposed study provides a review of the current state-of-the-art in deep learning for breast cancer detection and prognosis, including the use of convolutional neural networks (CNNs). To address the issue of breast cancer detection, a deep learning architectures named ResNet-50 and EffNet-50 are used to detect the affected portion. In this proposed method, the Dicom mammogram image is used for diagnosis. The designed architectures are tested with the training dataset of size 19,994 and a validation set of size 8,333 and its accuracy is found separately for both models and also by combining them. The combined ResNet50 and EffNet50 architectures are found to provide a highest accuracy of 98.6% whereas the individual models accuracy is 96.2% and 96.8% for ResNet-50 and EffNet-50 respectively