[4C3-J-13] AI application: diagnosis

[4C3-J-13-02] Subtype Classification of Breast Cancer based on MRI Image

〇Shioro Nakamura¹, Kenta Tanaka², Hajime Yokota⁴, Takuya Adachi⁵, Yoichi Machida⁵, Takuro Horikoshi⁴, Joji Ohta⁶, Yasukuni Mori³, Hiroki Suyari³ (1. Department of Informatics and Imaging Systems, Faculty of Engineering, Chiba University, 2. Graduate School of Science and Engineering, Chiba Univerrsity, 3. Graduate School of Science and Engineering, Chiba University., 4. Dept. Radiology, Chiba University Hospital, 5. Dept. Radiology, Kameda Medical Center, 6. Dept. Radiology, Chiba University Hospital.)

Keywords:Subtype Classification, Breast Cancer, MRI

Breast cancer can be roughly divided into 4 subtypes depending on the characteristics of genes. The treatment strategy of breast cancer depends on the subtypes, so it is necessary to determine the subtypes quickly and accurately. Currently, subtypes are classified by a method of collecting cancer cells from patients. This method causes a lot of pains on the patient. Therefore, in this paper, we propose a method to classify subtypes from MRI images of patients and aim to reduce the pains on patients. We use Residual Network, and gained an overall accuracy of 67.3%.