![[2602.14010] A Deployment-Friendly Foundational Framework for Efficient Computational Pathology](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.14010] A Deployment-Friendly Foundational Framework for Efficient Computational Pathology
Summary
This paper presents LitePath, a foundational framework for computational pathology that significantly reduces computational costs while maintaining high accuracy, enabling deployment on low-power hardware.
Why It Matters
The development of LitePath addresses the critical need for efficient computational pathology solutions that can operate on accessible hardware, enhancing clinical scalability and reducing the carbon footprint of AI deployments. This is particularly relevant as healthcare increasingly relies on AI for diagnostics.
Key Takeaways
- LitePath reduces model parameters by 28x and FLOPs by 403.5x compared to existing models.
- It processes 208 slides per hour on low-power devices, significantly improving efficiency.
- Achieves 99.71% of the AUC of the larger model Virchow2, demonstrating high accuracy.
- Introduces the Deployability Score (D-Score) to quantify the balance between accuracy and efficiency.
- Promotes energy-efficient pathology image analysis, reducing the carbon footprint of AI.
Computer Science > Computer Vision and Pattern Recognition arXiv:2602.14010 (cs) [Submitted on 15 Feb 2026] Title:A Deployment-Friendly Foundational Framework for Efficient Computational Pathology Authors:Yu Cai, Cheng Jin, Jiabo Ma, Fengtao Zhou, Yingxue Xu, Zhengrui Guo, Yihui Wang, Zhengyu Zhang, Ling Liang, Yonghao Tan, Pingcheng Dong, Du Cai, On Ki Tang, Chenglong Zhao, Xi Wang, Can Yang, Yali Xu, Jing Cui, Zhenhui Li, Ronald Cheong Kin Chan, Yueping Liu, Feng Gao, Xiuming Zhang, Li Liang, Hao Chen, Kwang-Ting Cheng View a PDF of the paper titled A Deployment-Friendly Foundational Framework for Efficient Computational Pathology, by Yu Cai and 25 other authors View PDF HTML (experimental) Abstract:Pathology foundation models (PFMs) have enabled robust generalization in computational pathology through large-scale datasets and expansive architectures, but their substantial computational cost, particularly for gigapixel whole slide images, limits clinical accessibility and scalability. Here, we present LitePath, a deployment-friendly foundational framework designed to mitigate model over-parameterization and patch level redundancy. LitePath integrates LiteFM, a compact model distilled from three large PFMs (Virchow2, H-Optimus-1 and UNI2) using 190 million patches, and the Adaptive Patch Selector (APS), a lightweight component for task-specific patch selection. The framework reduces model parameters by 28x and lowers FLOPs by 403.5x relative to Virchow2, enabling deployme...
