CellPrior-Net: Prior-Guided Nuclei Detection and Classification for H&E Whole-Slide Images

2026-07-01

AI Summary

This work addresses the challenges of nucleus detection and classification in H&E-stained whole-slide images, which arise from substantial morphological diversity, staining variations, imaging device heterogeneity, and artifacts. To tackle these issues, the authors propose CellPrior-Net, a novel framework that leverages the hematoxylin (H) channel as a prior to guide feature learning, incorporates a lightweight convolutional network to enhance nucleus-aware representation, and integrates a quality assessment module to establish an end-to-end nuclear quantification pipeline. Evaluated across eight multi-organ, multi-center datasets encompassing over 10.4 million nuclei, the method achieves accuracy on par with current state-of-the-art approaches while significantly improving inference speed and system scalability.

Abstract

Accurate nuclei detection and classification in hematoxylin and eosin (H and E) whole-slide images (WSIs) is a key task in computational pathology, particularly for quantitative analysis of the tumor microenvironment. However, this task remains highly challenging due to variations in nuclei morphology, staining procedures, scanners, organs, magnifications, and WSI artifacts. In addition, many existing pipelines rely on computationally demanding architectures and post-processing procedures, making gigapixel WSI analysis time consuming. In this work, CellPriorNet (CP Net) is proposed, an efficient nuclei detection and classification pipeline that utilizes a lightweight convolutional neural network architecture and hematoxylin (H) channel as prior information to enhance nuclei-aware feature learning. Extensive benchmarking was conducted against state of the art pipelines on 8 public and private datasets (total:10.4M nuclei) obtained from different organs, scanners, magnifications, and clinical centers. Experimental results demonstrate that CP Net achieves comparable performance while significantly reducing inference time. Furthermore, CellQuant Net was introduced, an end to end nuclei quantification pipeline, that integrates a quality assessment (QA) model to exclude regions with artifacts, followed by CP-Net cell detection and classification. The pipeline is publicly available on GitHub, and provides a potentially efficient and scalable framework for downstream computational pathology applications.

Problem

• nuclei detection
• nuclei classification
• H&E whole-slide images
• computational pathology
• WSI artifacts

Innovation

• prior-guided learning
• lightweight CNN
• nuclei detection and classification
• computational pathology
• whole-slide image analysis