Deep Learning for Joint Classification and Segmentation of Histopathology Imageopen access
- Authors
- Park, Hyun-Cheol; Ghimire, Raman; Poudel, Sahadev; Lee, Sang-Woong
- Issue Date
- Jul-2022
- Publisher
- LIBRARY & INFORMATION CENTER, NAT DONG HWA UNIV
- Keywords
- Histopathological image analysis; Whole-slide image; Segmentation; Classification; Patchbased method
- Citation
- JOURNAL OF INTERNET TECHNOLOGY, v.23, no.4, pp.903 - 910
- Journal Title
- JOURNAL OF INTERNET TECHNOLOGY
- Volume
- 23
- Number
- 4
- Start Page
- 903
- End Page
- 910
- URI
- https://scholarworks.bwise.kr/gachon/handle/2020.sw.gachon/86122
- DOI
- 10.53106/160792642022072304025
- ISSN
- 1607-9264
- Abstract
- Liver cancer is one of the most prevalent cancer deaths worldwide. Thus, early detection and diagnosis of possible liver cancer help in reducing cancer death. Histopathological Image Analysis (HIA) used to be carried out traditionally, but these are time-consuming and require expert knowledge. We propose a patch-based deep learning method for liver cell classification and segmentation. In this work, a two-step approach for the classification and segmentation of whole-slide image (WSI) is proposed. Since WSIs are too large to be fed into convolutional neural networks (CNN) directly, we first extract patches from them. The patches are fed into a modified version of U-Net with its equivalent mask for precise segmentation. In classification tasks, the WSIs are scaled 4 times, 16 times, and 64 times respectively. Patches extracted from each scale are then fed into the convolutional network with its corresponding label. During inference, we perform majority voting on the result obtained from the convolutional network. The proposed method has demonstrated better results in both classification and segmentation of liver cancer cells.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - IT융합대학 > 소프트웨어학과 > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/gachon/handle/2020.sw.gachon/86122)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.