UNIStainNet: Foundation-Model-Guided Virtual Staining of H&E to IHC

TL;DR

UNIStainNet is a novel foundation-model-guided deep learning approach that enables accurate virtual IHC staining from H&E images, improving efficiency and multi-marker capability in pathology diagnostics.

Contribution

It introduces UNIStainNet, a SPADE-UNet conditioned on pathology foundation model tokens, enabling multi-marker virtual staining with tissue-level semantic guidance and improved accuracy.

Findings

Achieves state-of-the-art metrics on four stains from a single model.

Performs best on multiple datasets, including BCI.

Remaining errors are systematic in non-tumor tissue.

Abstract

Virtual immunohistochemistry (IHC) staining from hematoxylin and eosin (H&E) images can accelerate diagnostics by providing preliminary molecular insight directly from routine sections, reducing the need for repeat sectioning when tissue is limited. Existing methods improve realism through contrastive objectives, prototype matching, or domain alignment, yet the generator itself receives no direct guidance from pathology foundation models. We present UNIStainNet, a SPADE-UNet conditioned on dense spatial tokens from a frozen pathology foundation model (UNI), providing tissue-level semantic guidance for stain translation. A misalignment-aware loss suite preserves stain quantification accuracy, and learned stain embeddings enable a single model to serve multiple IHC markers simultaneously. On MIST, UNIStainNet achieves state-of-the-art distributional metrics on all four stains (HER2, Ki67, ER, PR) from a single unified model, where prior methods typically train separate per-stain models. On BCI, it also achieves the best distributional metrics. A tissue-type stratified failure analysis reveals that remaining errors are systematic, concentrating in non-tumor tissue. Code is available at https://github.com/facevoid/UNIStainNet.