Research Article
Automatic Diagnosis of Breast Cancer from Histopathological Images Using Deep Learning Technique
@INPROCEEDINGS{10.1007/978-3-030-93709-6_42, author={Elbetel Taye Zewde and Gizeaddis Lamesgin Simegn}, title={Automatic Diagnosis of Breast Cancer from Histopathological Images Using Deep Learning Technique}, proceedings={Advances of Science and Technology. 9th EAI International Conference, ICAST 2021, Hybrid Event, Bahir Dar, Ethiopia, August 27--29, 2021, Proceedings, Part I}, proceedings_a={ICAST}, year={2022}, month={1}, keywords={Breast cancer Cancer sub-type Classification Grading ResNet Transfer learning}, doi={10.1007/978-3-030-93709-6_42} }- Elbetel Taye Zewde
Gizeaddis Lamesgin Simegn
Year: 2022
Automatic Diagnosis of Breast Cancer from Histopathological Images Using Deep Learning Technique
ICAST
Springer
DOI: 10.1007/978-3-030-93709-6_42
Abstract
Breast cancer is the primary cause of women cancer death globally. Advancement in screening methods and early diagnosis can increase survival from breast cancer. Clinical breast examination, imaging, and pathological assessment are common techniques of a breast cancer screening. Biopsy test is the standard breast cancer screening method due to its ability to identify types and sub-types of cancer. However, current diagnosis using this method is generally made by visual inspection. The manual technique is time taking, dreary, and subjective, that can also lead to misdiagnosis. The current article proposes an automatic diagnosis system for breast cancer based on the deep learning neural network model. The model was trained and validated on histopathological images obtained from online data sets and local data obtained from Jimma University Medical Center using a digital camera mounted on a microscope. All images were pre-processed and enhanced before being fed into the previously trained ResNet 50 model. The developed technique is able to classify breast cancer into benign and malignant and to their subtypes. The results of our test showed that the proposed technique is 96.75%, 96.7% and 95.78% for the benign subtype and the malignant subtype classification, respectively. The developed technique has a potential to be used as a computer aided diagnosis system for clinicians, particularly in low resources setting, where both resources and experience are limited.
