Breast Cancer Data Classification Using Xception-Based Neural Network

Breast cancer is accounted as the fifth leading cause of cancer deaths among females all around the world. These rising curves of morbidities and mortalities due to breast cancer demand the correct prognosis and early detection of disease. In this study, deep learning techniques have been used due to their faster and accurate estimation over machine learning techniques for image dataset and wider application areas. A novel methodology has been proposed for the classification of histopathological images in benign and malignant classes. The Xception-based CNN model with depth-wise separable architecture has been implemented. The combination of layer allows the model to converge at faster rate, avoid overfitting and produces results with better accuracy. The desired features have been extracted using augmentation techniques, and the model has trained using one cycle fine tuning. The performance of the model was evaluated using precision, accuracy, recall and F1 score. The proposed model gives high accuracy and outperformed the studies performed on similar datasets and samples. © 2023, The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd.