U-Net Architecture Application to Nuclei Cell Detection

摘要

In this paper U-Net neural network architecture and its’ variations for biomedical image processing is described. The application of the U-Net to the task of nuclei cell detection in particular is analyzed. The features and difficulties of the task are addressed. Several techniques to improve training and performance of the network are studied and, as a result, the model with specific parameters for the task of nuclei cell detection is proposed.