Few-Shot Learning for Annotation-Efficient Nucleus Instance Segmentation
Yu Ming, Zihao Wu, Jie Yang, Danyi Li, Yuan Gao, Changxin Gao, Gui-Song Xia, Yuanqing Li, Li Liang, Jin-Gang Yu
TL;DR
This work tackles the annotation bottleneck in nucleus instance segmentation by reframing the task as a generalized few-shot learning problem. It introduces SGFSIS, a four-branch network guided by a Guided Classification Module and Structural Guidance Modules, and uses a three-stage training pipeline (pre-training, meta-training, fine-tuning) to leverage a fully annotated external dataset while handling overlapping base/novel classes. The approach achieves competitive performance with far fewer annotations (less than 5% of fully supervised labels) and shows particular strength in AJI by transferring cross-dataset structural knowledge. This framework enables annotation-efficient nucleus segmentation across diverse public datasets and provides a pathway to robust performance when labeled data on the target domain are scarce.
Abstract
Nucleus instance segmentation from histopathology images suffers from the extremely laborious and expert-dependent annotation of nucleus instances. As a promising solution to this task, annotation-efficient deep learning paradigms have recently attracted much research interest, such as weakly-/semi-supervised learning, generative adversarial learning, etc. In this paper, we propose to formulate annotation-efficient nucleus instance segmentation from the perspective of few-shot learning (FSL). Our work was motivated by that, with the prosperity of computational pathology, an increasing number of fully-annotated datasets are publicly accessible, and we hope to leverage these external datasets to assist nucleus instance segmentation on the target dataset which only has very limited annotation. To achieve this goal, we adopt the meta-learning based FSL paradigm, which however has to be tailored in two substantial aspects before adapting to our task. First, since the novel classes may be inconsistent with those of the external dataset, we extend the basic definition of few-shot instance segmentation (FSIS) to generalized few-shot instance segmentation (GFSIS). Second, to cope with the intrinsic challenges of nucleus segmentation, including touching between adjacent cells, cellular heterogeneity, etc., we further introduce a structural guidance mechanism into the GFSIS network, finally leading to a unified Structurally-Guided Generalized Few-Shot Instance Segmentation (SGFSIS) framework. Extensive experiments on a couple of publicly accessible datasets demonstrate that, SGFSIS can outperform other annotation-efficient learning baselines, including semi-supervised learning, simple transfer learning, etc., with comparable performance to fully supervised learning with less than 5% annotations.