Penggunaan Augmentasi Data pada Klasifikasi Jenis Kanker Payudara dengan Model Resnet-34

Abstract

A recent study from the Global Cancer Observatory (GLOBOCAN) revealed that in 2020 about 2.2 milion women worldwide have been diagnosed with breast cancer. Diagnostic tissue biopsy with hematoxylin and eosin stained images is used to make decisions on the final diagnosis. Computer-assisted diagnostic systems contribute to increasing the efficiency of this process. In this study using the dataset “BreAst Cancer Histology Images (BACH)”. And a method was made to classify breast biopsy images stained with hematoxylin and eosin using convolutional neural networks. The images are classified into four classes, normal tissue, benign lesions, carcinoma in situ and invasive carcinoma. In this study, regularization techniques were also carried out in the convolutional neural networks model to achieve maximum accuracy.

Keywords— Breast Cancer, Convolutional Neural Networks, Data Augmentation, Multiclass Classification.