Breast Cancer Classification of Histopathological Images using Deep Convolutional Neural Networks
Athanasios Kanavos, Efstratios Kolovos, Orestis Papadimitriou, Manolis Μaragoudakis
Abstract
Histopathology refers to the diagnosis of tissue diseases and involves the thorough examination of tissues and cells under a microscope. Tissues are collected by biopsy and viewed under the microscope after being properly processed. Modern medical image processing techniques involve the collection of histopathological images taken under a microscope and their analysis using different algorithms and techniques. Deep Learning is widely used in the field of medical imaging as it does not require any specialized prior knowledge in the problem domain. The dataset used for our experiments comprises of histopathological scans derived from the PatchCamelyon dataset. Various Convolutional Neural Network architectures were implemented, where their hyperparameters were fine tuned and the classification results are presented. The deep learning neural networks are accessed for their worth in terms of accuracy, loss, AUC, precision, recall and time required.