HaDM-ST: Histology-Assisted Differential Modeling for Spatial Transcriptomics Generation

TY - GEN

T1 - HaDM-ST

T2 - 4th International Workshop on Computational Mathematics Modeling in Cancer Analysis, CMMCA 2025, Held in Conjunction with the 28th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2025

AU - Liu, Xuepeng

AU - Jiang, Zheng

AU - Zhu, Pinan

AU - Liu, Hanyu

AU - Li, Chao

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

PY - 2026

Y1 - 2026

N2 - 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 (HR) ST generation framework conditioned on H&E images and low-resolution (LR) 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-res 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 HR ST predictions.

AB - 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 (HR) ST generation framework conditioned on H&E images and low-resolution (LR) 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-res 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 HR ST predictions.

KW - Diffusion Models

KW - Gene Expression Prediction

KW - Histology-to-Transcriptomics Translation

KW - Spatial Transcriptomics

UR - https://www.scopus.com/pages/publications/105018802078

U2 - 10.1007/978-3-032-06624-4_14

DO - 10.1007/978-3-032-06624-4_14

M3 - Conference contribution

AN - SCOPUS:105018802078

SN - 9783032066237

VL - 16178

T3 - Lecture Notes in Computer Science

SP - 129

EP - 138

BT - Computational Mathematics Modeling in Cancer Analysis - 4th International Workshop, CMMCA 2025, Held in Conjunction with MICCAI 2025, Proceedings

A2 - Li, Chao

A2 - Qin, Wenjian

A2 - Wu, Jia

A2 - Zaki, Nazar

PB - Springer Science and Business Media Deutschland GmbH

CY - Switzerland

Y2 - 27 September 2025 through 27 September 2025

ER -