Nuclei R-CNN: Improve Mask R-CNN for Nuclei Segmentation

Accurate nuclei segmentation plays an essential role in medical research and various clinical applications. Recently, deep learning has demonstrated its superior performance on object segmentation in natural scene images. However, these methods cannot produce fine segmentation in histopathological images. Therefore, this paper improves Mask R-CNN for nuclei segmentation which is called Nuclei R-CNN, mainly focus on the network model, training scheme and the preprocess of histopathological data. Additionally, we found poor prediction accuracy under high-resolution histopathological images and proposed a simple Block-Merging method to deal with it. The quantitative assessment of Nuclei R-CNN has been based on challenging public dataset. Our experimental shows that Nuclei R-CNN outperforms existing methods for segmenting histopathological images.