MINT: Molecularly Informed Training with Spatial Transcriptomics Supervision for Pathology Foundation Models
This work provides a method to incorporate molecular information into pathology foundation models, which is significant for researchers and clinicians seeking a more comprehensive understanding of tissue states beyond morphology.
This paper introduces MINT, a fine-tuning framework that integrates spatial transcriptomics supervision into pretrained pathology Vision Transformers. MINT achieves a mean Pearson r of 0.440 for gene expression prediction on HEST-Bench and an EVA score of 0.803 for general pathology tasks, outperforming existing methods.
Pathology foundation models learn morphological representations through self-supervised pretraining on large-scale whole-slide images, yet they do not explicitly capture the underlying molecular state of the tissue. Spatial transcriptomics technologies bridge this gap by measuring gene expression in situ, offering a natural cross-modal supervisory signal. We propose MINT (Molecularly Informed Training), a fine-tuning framework that incorporates spatial transcriptomics supervision into pretrained pathology Vision Transformers. MINT appends a learnable ST token to the ViT input to encode transcriptomic information separately from the morphological CLS token, preventing catastrophic forgetting through DINO self-distillation and explicit feature anchoring to the frozen pretrained encoder. Gene expression regression at both spot-level (Visium) and patch-level (Xenium) resolutions provides complementary supervision across spatial scales. Trained on 577 publicly available HEST samples, MINT achieves the best overall performance on both HEST-Bench for gene expression prediction (mean Pearson r = 0.440) and EVA for general pathology tasks (0.803), demonstrating that spatial transcriptomics supervision complements morphology-centric self-supervised pretraining.