Domain-incremental White Blood Cell Classification with Privacy-aware Continual Learning

2025-03-25

AI Summary

White blood cell (WBC) classification is critical for clinical diagnosis, yet model generalization degrades significantly due to domain shift—arising from heterogeneous sample sources (peripheral blood vs. bone marrow) and imaging conditions—and catastrophic forgetting during continual learning. To address these challenges, we propose a privacy-preserving generative replay continual learning framework tailored for dynamic clinical environments. Our approach uniquely integrates a lightweight generator into medical vision foundation models (ResNet50, RetCCL, CTransPath, UNI), enabling implicit feature synthesis for cross-domain incremental adaptation while preserving historical task performance. Evaluated on four real-world, multi-source WBC datasets across four cross-domain task sequences, our method achieves an average accuracy gain of 12.6% over standard fine-tuning, with near-zero backward transfer degradation. These results demonstrate strong deployment feasibility and robustness in practical healthcare settings.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems

Abstract

White blood cell (WBC) classification plays a vital role in hematology for diagnosing various medical conditions. However, it faces significant challenges due to domain shifts caused by variations in sample sources (e.g., blood or bone marrow) and differing imaging conditions across hospitals. Traditional deep learning models often suffer from catastrophic forgetting in such dynamic environments, while foundation models, though generally robust, experience performance degradation when the distribution of inference data differs from that of the training data. To address these challenges, we propose a generative replay-based Continual Learning (CL) strategy designed to prevent forgetting in foundation models for WBC classification. Our method employs lightweight generators to mimic past data with a synthetic latent representation to enable privacy-preserving replay. To showcase the effectiveness, we carry out extensive experiments with a total of four datasets with different task ordering and four backbone models including ResNet50, RetCCL, CTransPath, and UNI. Experimental results demonstrate that conventional fine-tuning methods degrade performance on previously learned tasks and struggle with domain shifts. In contrast, our continual learning strategy effectively mitigates catastrophic forgetting, preserving model performance across varying domains. This work presents a practical solution for maintaining reliable WBC classification in real-world clinical settings, where data distributions frequently evolve.

Problem

• Address domain shifts in WBC classification across hospitals
• Prevent catastrophic forgetting in dynamic medical data environments
• Enable privacy-preserving continual learning for foundation models

Innovation

• Generative replay-based continual learning strategy
• Privacy-preserving synthetic latent representation
• Lightweight generators mimic past data