KLASIFIKASI INVASIVE DUCTAL CARCINOMA MENGGUNAKAN CONVOLUTIONAL NEURAL NETWORK

Breast cancer is one of the top cancers affecting women in the world. According to WHO, it is estimated worldwide that more than 508,000 women have died in 2011 from breast cancer. Invasive ductal carcinoma is a type of breast cancer that often occurs in women in the United States. Research in detecting IDC from histopathological images using CNN is still limited. Therefore, this study proposes a method with the fine tune model of MobileNet and VGG16 to detect IDC from histopathological images and compare between the two models to get better accuracy. This result has an increase in CNN performance in the IDC classification. The proposed model achieved an f1-score of 0.87 on MobileNet and 0.88 on VGG16 with an increase in f-1 score and balanced accuracy reaching 1.72% and 1.21% on MobileNet, as well as an increase in f-1 score and balanced accuracy reaching 2.72% and 2.49% in VGG16 from previous research