HaDM-ST: Histology-Assisted Differential Modeling for Spatial Transcriptomics Generation
This work improves spatial transcriptomics resolution for biomedical researchers, though it is incremental as it builds on existing diffusion-based frameworks.
The paper tackled the problem of generating high-resolution spatial transcriptomics from low-resolution data and H&E histology images by addressing challenges in feature extraction, multimodal alignment, and gene-specific modeling, resulting in HaDM-ST outperforming prior methods across 200 genes in diverse tissues and species.
Spatial transcriptomics (ST) reveals spatial heterogeneity of gene expression, yet its resolution is limited by current platforms. Recent methods enhance resolution via H&E-stained histology, but three major challenges persist: (1) isolating expression-relevant features from visually complex H&E images; (2) achieving spatially precise multimodal alignment in diffusion-based frameworks; and (3) modeling gene-specific variation across expression channels. We propose HaDM-ST (Histology-assisted Differential Modeling for ST Generation), a high-resolution ST generation framework conditioned on H&E images and low-resolution ST. HaDM-ST includes: (i) a semantic distillation network to extract predictive cues from H&E; (ii) a spatial alignment module enforcing pixel-wise correspondence with low-resolution ST; and (iii) a channel-aware adversarial learner for fine-grained gene-level modeling. Experiments on 200 genes across diverse tissues and species show HaDM-ST consistently outperforms prior methods, enhancing spatial fidelity and gene-level coherence in high-resolution ST predictions.