Multi-Path Dilated Residual Network for Nuclei Segmentation and Detection.

Eric Ke Wang,Leyun Pan,Caixia Cheng,Antonia Dimitrakopoulou-Strauss,Nie Zhe,Yueping Li,Xun Zhang

doi:10.3390/cells8050499

Abstract

As a typical biomedical detection task, nuclei detection has been widely used in human health management, disease diagnosis and other fields. However, the task of cell detection in microscopic images is still challenging because the nuclei are commonly small and dense with many overlapping nuclei in the images. In order to detect nuclei, the most important key step is to segment the cell targets accurately. Based on Mask RCNN model, we designed a multi-path dilated residual network, and realized a network structure to segment and detect dense small objects, and effectively solved the problem of information loss of small objects in deep neural network. The experimental results on two typical nuclear segmentation data sets show that our model has better recognition and segmentation capability for dense small targets.

Highlights

Segmentation and detection of nuclei in microscopic histopathological images is of great significance
Because we find that the instance segmentation model based on dilated residual network can effectively solve the problem of small object information loss in the object detection tasks, and the model has the function of balancing the receptive field of deep network and the scale of feature map
We evaluate the effect of Mask RCNN

Summary

Introduction

Segmentation and detection of nuclei in microscopic histopathological images is of great significance. It is important to determine the morphology and location of nuclei in some cancer pathological images for grading diagnosis of malignant and benign cancers. In clinical practice, renal carcinoma cells are often arranged in sheet, strip, acinar or tubular shape, much like renal tubules. Employing computer aided image analysis technology to effectively and accurately segment the nuclei in cancer pathological images is an urgent need to analyze the malignant degree of renal cancer and build an automatic classification system for renal cancer. In addition to the nuclear segmentation of cancer pathological images, other nuclear segmentation has value, for instance, clinical diagnosis of hematopoietic diseases can be done by observing the number, proportion and morphological changes of different types of white blood cells

Methods

Results

Conclusion