A Hybrid-Attention Nested UNet for Nuclear Segmentation in Histopathological Images.

Hongliang He,Yongjie Xu,Jihua Wu,Yongsheng Liang,Luyang Chen,Jie Chen,Chi Zhang,Yanchang Lu,Ruizhe Geng

doi:10.3389/fmolb.2021.614174

Abstract

Nuclear segmentation of histopathological images is a crucial step in computer-aided image analysis. There are complex, diverse, dense, and even overlapping nuclei in these histopathological images, leading to a challenging task of nuclear segmentation. To overcome this challenge, this paper proposes a hybrid-attention nested UNet (Han-Net), which consists of two modules: a hybrid nested U-shaped network (H-part) and a hybrid attention block (A-part). H-part combines a nested multi-depth U-shaped network and a dense network with full resolution to capture more effective features. A-part is used to explore attention information and build correlations between different pixels. With these two modules, Han-Net extracts discriminative features, which effectively segment the boundaries of not only complex and diverse nuclei but also small and dense nuclei. The comparison in a publicly available multi-organ dataset shows that the proposed model achieves the state-of-the-art performance compared to other models.

Highlights

Histopathological imaging diagnosis is an important significance of cancer diagnosis, known as the “gold standard” of clinical tumors
To address the issues mentioned above, the main contributions of this paper are 1) We propose a hybridattention nested UNet (Han-Net), which consists of two modules: a hybrid nested U-shaped network (H-part) and hybrid attention block (A-part)
We validate the performance of our proposed model in the Multi-Organ nuclear segmentation (MoNuSeg) dataset (Kumar et al, 2017)

Summary

Introduction

Histopathological imaging diagnosis is an important significance of cancer diagnosis, known as the “gold standard” of clinical tumors. Nuclear segmentation of histopathological images is a crucial step in computer-aided image analysis. Segmenting the nucleus in pathological tissue sections provides powerful support for disease diagnosis, cancer staging, and postoperative treatment. Traditional nuclear segmentation methods have contributed to some extent, such as Otsu (Otsu, 1979), the watershed method (Yang et al, 2006), K-mean clustering (Filipczuk et al, 2011), and Grab Cut (Rother et al, 2004). Some specific parameters or thresholds are required to set while using these methods for nuclear segmentation. The lack of generalization ability makes these methods only effective for a few types of histopathological images. With the application and development of deep learning technology in image segmentation, these traditional nuclear segmentation methods are only used as pre/post-processing steps

Methods

Results

Conclusion