DARC: Distribution-Aware Re-Coloring Model for Generalizable Nucleus Segmentation

TL;DR

This paper introduces DARC, a novel model for nucleus segmentation that addresses domain gaps caused by staining variations and foreground-background ratio differences, improving robustness across diverse pathological image datasets.

Contribution

The paper proposes a distribution-aware re-coloring method and a new instance normalization technique to enhance generalization in nucleus segmentation tasks.

Findings

DARC improves segmentation accuracy across multiple datasets.

The re-coloring method reduces color variation effects.

The normalization technique handles foreground-background ratio differences.

Abstract

Nucleus segmentation is usually the first step in pathological image analysis tasks. Generalizable nucleus segmentation refers to the problem of training a segmentation model that is robust to domain gaps between the source and target domains. The domain gaps are usually believed to be caused by the varied image acquisition conditions, e.g., different scanners, tissues, or staining protocols. In this paper, we argue that domain gaps can also be caused by different foreground (nucleus)-background ratios, as this ratio significantly affects feature statistics that are critical to normalization layers. We propose a Distribution-Aware Re-Coloring (DARC) model that handles the above challenges from two perspectives. First, we introduce a re-coloring method that relieves dramatic image color variations between different domains. Second, we propose a new instance normalization method that is robust to the variation in foreground-background ratios. We evaluate the proposed methods on two H$\&$E stained image datasets, named CoNSeP and CPM17, and two IHC stained image datasets, called DeepLIIF and BC-DeepLIIF. Extensive experimental results justify the effectiveness of our proposed DARC model. Codes are available at \url{https://github.com/csccsccsccsc/DARC