Radiotekhnika
Publishing house Radiotekhnika

"Publishing house Radiotekhnika":
scientific and technical literature.
Books and journals of publishing houses: IPRZHR, RS-PRESS, SCIENCE-PRESS


Тел.: +7 (495) 625-9241

 

Neural network approach to cell segmentation in immunocytochemical study

DOI 10.18127/j15604136-201805-10

Keywords:

Dmitry Parpulov -  Bauman Moscow State Technical University, Moscow, Russian Federation

Andrey Samorodov -  Bauman Moscow State Technical University, Moscow, Russian Federation

Vladimir Iglovikov -  Lyft Inc., San Francisco, CA 94107, USA


Immunocytochemical (ICH) study has a great importance in the evaluation of some predictive and prognostic factors in breast cancer (BC). The degree of C-erbB-2 oncoprotein expression is important for BC treatment strategy choice, because C-erbB-2 is a receptor for HER2/neu epidermal growth factor [6]. Overexpression of this oncoprotein is a risk factor of BC recurrence. The degree of membrane staining (qualitatively from 0 till 3+) represents tumor HER2 status in ICH study.
Now tumor HER2 status is evaluated manually by oncologist, doctor should check about 100-200 images of the specimen before giving the final assessment. Thus, the automation of determining tumor HER2 status is an urgent task, because it will allow to free an oncologist from routine work.
Automatization of tumor HER2 status assessment is divided into 2 subtasks: cells segmentation and automatic determining the degree of membrane staining. This work is devoted to the problem of cell segmentation. For classical methods of computer vision the problem of cell segmentation is difficult due to the frequent presence of erythrocytic background and non-cellular elements at specimen image. In addition, the process of feature engineering for these algorithms is a time consuming and nontrivial. We propose segmentation algorithm, based on deep convolutional neural networks (CNN), which are very popular in image processing [1,5]. Using this approach, we got better results, than using classical computer vision algorithms.

References:
  1. Ching, T., Himmelstein, D.S., Beaulieu-Jones, B.K., Kalinin, A.A., Do, B.T., Way, G.P., Ferrero, E., Agapow, P.M., Xie, W., Rosen, G.L., et al.: Opportunities and obstacles for deep learning in biology and medicine. bioRxiv p. 142760 (2017)
  2. J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li and L. Fei-Fei, ImageNet: A Large-Scale Hierarchical Image Database. IE- EE Computer Vision and Pattern Recognition (CVPR), 2009
  3. Kingma, D., Ba, J.: Adam: A method for stochastic opti- mization. arXiv preprint arXiv:1412.6980 (2014)
  4. C. J. De Luca. The use of surface electromyography in biome- chanics. J Appl Biomech, 13(2):135–163, 1997.
  5. Ronneberger, O., Fischer, P., Brox, T.: U-net: Convolutional networks for biomedical image segmentation. In: Interna- tional Conference on Medical Image Computing and Com- puter-Assisted Intervention. pp. 234–241. Springer (2015)
  6. Simonyan, K., Zisserman, A.: Very deep convolutional net- works for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014)
  7. Volchenko N.N., Slavnova E.N., Gladunova Z.D. and others. Current cytological diagnostics of breast diseases. Moscow: Publishing house of Bauman Moscow State Technical Uni- versity, 2014

© Издательство «РАДИОТЕХНИКА», 2004-2017            Тел.: (495) 625-9241                   Designed by [SWAP]Studio