Cloud storage based diagnosis of breast cancer using novel transfer learning with multi-layer perceptron

Early breast cancer detection is vital to enhance the human survival rate. Hence, this study aimed to segment cancer affected areas using Transfer Learning based Cancer Segmentation (TL-CAN-Seg) methodology. It also intended to extract the relevant features and then store them in the cloud. In addition, it introduced a novel MLP (Multi-layer perceptron) with tuned Levenberg Marquadrt algorithm (LM) to effectively classify the breast cancer affected-area. Initially, the mammogram image taken and image pre-processing is performed. Subsequently, segmentation is carried out using novel TL-CAN-Seg and feature extraction is performed, stored in cloud. Finally, classification is performed to classify the breast cancer. The proposed MLP with LM has the ability to converge faster and the proposed TL-CAN-Seg also has ability to learn the complex image patterns that eventually increases accuracy of breast cancer diagnosis than the traditional methods. This is confirmed through the comparative analysis in the results.