MIPHEI-ViT v1.0

MIPHEI-ViT v1.0

This release provides the minimal components required to run MIPHEI-ViT, a model that predicts 16-channel multiplex immunofluorescence (mIF) from H&E histology images using a ViT-based encoder and U-Net decoder.

Included files

• model.safetensors — main model weights
• model.py — minimal PyTorch implementation for loading and inference
• config.yaml — model configuration and metadata
• logreg.pth — logistic regression head for OrionCRC cell-type prediction
• requirements.txt — minimal inference dependencies
• LICENSE — license information

Notes

This package provides inference-only code.
Training code is not included in this release.

Citation

Citation

Balezo et al., “MIPHEI-ViT: Multiplex Immunofluorescence Prediction from H&E Images using ViT Foundation Models,”
arXiv:2505.10294 (2025).