CountXplain: Interpretable Cell Counting with Prototype-Based Density Map Estimation

2025-11-24

AI Summary

Biomedical cell counting faces challenges in deep learning model interpretability. This paper introduces prototype learning into density map estimation for the first time, proposing an interpretable cell counting framework: a prototype layer is embedded within the density estimation network to automatically learn biologically meaningful visual prototypes representing cells and background artifacts; cell regions are identified via prototype matching, and pixel-level interpretation heatmaps are generated using gradient-weighted class activation mapping. The method achieves competitive counting accuracy (MAE/ME) against state-of-the-art models on two public benchmarks while providing intuitive, biologist-validated visual explanations. Key contributions include: (i) the first prototype-based interpretable framework specifically designed for cell counting; (ii) joint optimization of counting performance and model interpretability; and (iii) significantly enhanced clinical trustworthiness and transparency for computer-aided diagnosis.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Machine Learning: Calibration & Uncertainty Quantification
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics

Abstract

Cell counting in biomedical imaging is pivotal for various clinical applications, yet the interpretability of deep learning models in this domain remains a significant challenge. We propose a novel prototype-based method for interpretable cell counting via density map estimation. Our approach integrates a prototype layer into the density estimation network, enabling the model to learn representative visual patterns for both cells and background artifacts. The learned prototypes were evaluated through a survey of biologists, who confirmed the relevance of the visual patterns identified, further validating the interpretability of the model. By generating interpretations that highlight regions in the input image most similar to each prototype, our method offers a clear understanding of how the model identifies and counts cells. Extensive experiments on two public datasets demonstrate that our method achieves interpretability without compromising counting effectiveness. This work provides researchers and clinicians with a transparent and reliable tool for cell counting, potentially increasing trust and accelerating the adoption of deep learning in critical biomedical applications. Code is available at https://github.com/NRT-D4/CountXplain.

Problem

• Addressing interpretability challenges in deep learning cell counting models
• Developing prototype-based density map estimation for transparent cell counting
• Providing reliable biomedical counting tools without compromising effectiveness

Innovation

• Prototype-based density map estimation for cell counting
• Learns representative visual patterns for cells and background
• Generates interpretations highlighting prototype-similar regions in images