WSI-INR: Implicit Neural Representations for Lesion Segmentation in Whole-Slide Images

2026-03-04

AI Summary

This work proposes WSI-INR, the first patch-free framework leveraging implicit neural representations (INRs) for pathological lesion segmentation in whole-slide images (WSIs). Traditional methods rely on discrete image patches, disrupting spatial continuity and treating multi-resolution views as independent samples, which leads to fragmented segmentation and sensitivity to resolution changes. In contrast, WSI-INR models different resolutions as varying sampling densities of a single continuous tissue representation via multi-resolution hash-grid encoding and employs a shared decoder to learn a universal cross-case prior. This approach preserves global spatial coherence while enabling end-to-end joint training, achieving a 26.11% improvement in Dice score at Base/4 resolution—substantially outperforming U-Net (↓54.28%) and TransUNet (↓36.18%)—and significantly enhancing robustness across resolutions.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Representation Learning
• Knowledge Representation and Reasoning: Argumentation

Application Category

• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Whole-slide images (WSIs) are fundamental for computational pathology, where accurate lesion segmentation is critical for clinical decision making. Existing methods partition WSIs into discrete patches, disrupting spatial continuity and treating multi-resolution views as independent samples, which leads to spatially fragmented segmentation and reduced robustness to resolution variations. To address the issues, we propose WSI-INR, a novel patch-free framework based on Implicit Neural Representations (INRs). WSI-INR models the WSI as a continuous implicit function mapping spatial coordinates directly to tissue semantics features, outputting segmentation results while preserving intrinsic spatial information across the entire slide. In the WSI-INR, we incorporate multi-resolution hash grid encoding to regard different resolution levels as varying sampling densities of the same continuous tissue, achieving a consistent feature representation across resolutions. In addition, by jointly training a shared INR decoder, WSI-INR can capture general priors across different cases. Experimental results showed that WSI-INR maintains robust segmentation performance across resolutions; at Base/4, our resolution-specific optimization improves Dice score by +26.11%, while U-Net and TransUNet decrease by 54.28% and 36.18%, respectively. Crucially, this work enables INRs to segment highly heterogeneous pathological lesions beyond structurally consistent anatomical tissues, offering a fresh perspective for pathological analysis.

Problem

• whole-slide images
• lesion segmentation
• spatial continuity
• resolution variation
• computational pathology

Innovation

• Implicit Neural Representations
• Whole-Slide Image
• Lesion Segmentation
• Multi-resolution Hash Encoding
• Patch-free Framework