UN-SAM: Universal Prompt-Free Segmentation for Generalized Nuclei Images

Zhen Chen,Qing Xu,Xinyu Liu,Yixuan Yuan

2024-02-26

Abstract:In digital pathology, precise nuclei segmentation is pivotal yet challenged by the diversity of tissue types, staining protocols, and imaging conditions. Recently, the segment anything model (SAM) revealed overwhelming performance in natural scenarios and impressive adaptation to medical imaging. Despite these advantages, the reliance of labor-intensive manual annotation as segmentation prompts severely hinders their clinical applicability, especially for nuclei image analysis containing massive cells where dense manual prompts are impractical. To overcome the limitations of current SAM methods while retaining the advantages, we propose the Universal prompt-free SAM framework for Nuclei segmentation (UN-SAM), by providing a fully automated solution with remarkable generalization capabilities. Specifically, to eliminate the labor-intensive requirement of per-nuclei annotations for prompt, we devise a multi-scale Self-Prompt Generation (SPGen) module to revolutionize clinical workflow by automatically generating high-quality mask hints to guide the segmentation tasks. Moreover, to unleash the generalization capability of SAM across a variety of nuclei images, we devise a Domain-adaptive Tuning Encoder (DT-Encoder) to seamlessly harmonize visual features with domain-common and domain-specific knowledge, and further devise a Domain Query-enhanced Decoder (DQ-Decoder) by leveraging learnable domain queries for segmentation decoding in different nuclei domains. Extensive experiments prove that UN-SAM with exceptional performance surpasses state-of-the-arts in nuclei instance and semantic segmentation, especially the generalization capability in zero-shot scenarios. The source code is available at

Image and Video Processing,Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The paper attempts to address the main challenges faced in the task of nuclear segmentation in digital pathology, including the segmentation difficulties caused by the diversity of tissue types, staining protocols, and imaging conditions. Specifically, existing segmentation methods perform well on natural images but face the following two major issues in medical image segmentation, especially nuclear image segmentation:

1. **Dependence on manual annotation**: Current segmentation models (such as SAM) require a large amount of manual annotation as prompts to guide the segmentation decoding, which is very time-consuming and impractical in clinical applications, especially in nuclear image analysis involving a large number of cells.

2. **Lack of generalization ability**: Existing medical SAM algorithms have limited generalization ability across different datasets, especially when faced with the diversity of tissue types, staining protocols, and imaging conditions, making it difficult to maintain high accuracy.

To address these issues, the paper proposes a universal prompt-free segmentation framework called UN-SAM (Universal Prompt-Free Segmentation for Generalized Nuclei Images), aiming to achieve automated nuclear instance and semantic segmentation with excellent generalization ability. Specifically, UN-SAM addresses the above issues through the following innovations:

- **Multi-Scale Self-Prompt Generation Module (SPGen)**: Automatically generates high-quality mask prompts, eliminating the need for manual annotation and simplifying clinical workflows.
- **Domain-Adaptive Tuning Encoder (DT-Encoder)**: Enhances visual features through domain-general and domain-specific knowledge, improving the model's generalization ability.
- **Domain Query Enhanced Decoder (DQ-Decoder)**: Utilizes learnable domain queries for segmentation decoding, further enhancing the segmentation performance of nuclear images from different domains.

Through these designs, UN-SAM demonstrates outstanding performance on multiple nuclear image datasets, particularly excelling in zero-shot scenarios, significantly outperforming existing medical segmentation methods in terms of generalization ability.