ContiStain: Cross-Domain Relation-Preserving Distillation for Continual Multi-Domain Virtual IHC Staining

2026-07-04

AI Summary

This work addresses catastrophic forgetting in continual learning for virtual immunohistochemistry (IHC) staining of novel biomarkers, where conventional fine-tuning disrupts the representational relationships among previously learned markers. To mitigate this, the authors propose a domain-aware mixture-of-experts (MoE) feature extraction architecture coupled with a cross-domain relational preservation distillation mechanism. This mechanism enforces consistency in token-level cosine similarity matrices across domains, thereby preserving the original representational structure while adapting to new tasks. Evaluated under a four-domain continual learning setup on the MIST dataset, the proposed method significantly outperforms sequential fine-tuning, reducing FID by 11.1 and ConchFID by 60.9, demonstrating enhanced model stability and scalability.

Abstract

A unified multiplex virtual staining model enables scalable and non-destructive multiplex analysis from H&E slides while promoting parameter efficiency, shared pathological knowledge, and consistent cross-biomarker representations. However, in clinical practice, data for new biomarkers are typically acquired sequentially over time. Fine-tuning on such temporally arriving data leads to severe performance degradation on previously learned biomarkers, as sequential optimization disrupts the structured relationships among biomarker representations in the latent space. To address this issue, we propose ContiStain, an IHC multi-domain relational distillation framework for continual virtual staining. We first (i) construct a domain-aware structured feature space using a mixture-of-experts (MoE) feature extractor to reduce representation interference across biomarker domains. Based on this stabilized feature space, we then (ii) propose a relation-preserving distillation strategy that explicitly enforces the consistency of cross-domain token-level cosine similarity matrices between learned biomarker domains during continual adaptation. By maintaining cross-domain structural coherence, ContiStain mitigates forgetting while retaining adaptability to new domains. Experiments on the MIST dataset under a four-domain sequential virtual IHC staining setting show improved stability, reducing FID and ConchFID by 11.1 and 60.9 compared to sequential fine-tuning, enabling scalable and robust multi-domain virtual staining. Code is released at https://github.com/ccitachi/ContiStain.

Problem

• continual learning
• virtual staining
• catastrophic forgetting
• cross-domain representation
• multi-domain IHC

Innovation

• continual learning
• virtual staining
• relation-preserving distillation
• mixture-of-experts
• cross-domain representation