Semantic Nuclei Segmentation with Deep Learning on Breast Pathology Images


TURAN S. , BİLGİN G.

International Scientific Meeting on Electrical-Electronics and Biomedical Engineering and Computer Science (EBBT), İstanbul, Türkiye, 24 - 26 Nisan 2019 identifier identifier

Özet

The methods of facilitating the workload of doctors are tried to be developed in the diagnosis of cancer. One of these procedures is the segmentation of cell nuclei in digital images obtained in the field of pathology. For the segmentation process, deep-learning can be performed with local small patches obtained from the digital image, and pixel-based systems can be developed by using semantic segmentation technique. In this study, histopathological images obtained from hematoxylin and eosin staining are used for biopsy samples taken for diagnosis of breast cancer. The studies were performed in Matlab environment by using SegNet and U-Net algorithms and the accuracy of semantic segmentation was evaluated comparatively.