HiFi-ST: Histology-Conditioned Local Coordinate Modeling for Spatial Gene Expression Prediction
Spatial transcriptomics (ST) enables the characterization of spatial gene expression patterns in tissues, but its discrete measurement locations limit the availability of spatially resolved molecular information. To address this challenge, we propose HiFi-ST, a histology-conditioned framework for spatial gene expression prediction based on local coordinate modeling. HiFi-ST integrates multiscale hematoxylin and eosin (H&E) image representations with spatial coordinate information. Multiple local coordinate queries are performed within each spot neighborhood, and their expression predictions are combined through stochastic aggregation to generate spot-level estimates under observed expression supervision. Experiments on three public datasets, HER2+, cSCC, and Alex_NatGen, achieved mean gene-wise Pearson correlation coefficients (PCC (ACG)) of 0.3547, 0.3314, and 0.3385, respectively. Compared with previously published benchmark reference results, HiFi-ST achieved higher overall correlation on HER2+ and Alex_NatGen, while obtaining a slightly higher PCC (ACG) and lower prediction errors on cSCC. Gene-level spatial expression analyses further demonstrated that the predicted expression preserved selected spatial patterns associated with tissue morphology. In addition, spatial-domain clustering across six HER2+ sections achieved a mean adjusted Rand index of 0.3610 and a mean normalized mutual information of 0.3847, indicating the preservation of spatial information relevant to tissue-region organization. Overall, HiFi-ST provides an effective approach to histology-driven spatial gene expression prediction by integrating multiscale morphological features, local coordinate modeling, and spot-level aggregation.
References
- Ståhl, P. L., Salmén, F., Vickovic, S., et al. (2016). Visualization and analysis of gene expression in tissue sections by spatial transcriptomics. Science, 353(6294), 78–82. https://doi.org/10.1126/science.aaf2403.
- Burgess, D. J. (2019). Spatial transcriptomics coming of age. Nature Reviews Genetics, 20(6), 317. https://doi.org/10.1038/s41576-019-0133-8.
- Pang, M., Su, K., & Li, M. (2021). Leveraging information in spatial transcriptomics to predict super-resolution gene expression from histology images in tumors. bioRxiv. https://doi.org/10.1101/2021.11.28.470212.
- Zeng, Y., Wei, Z., Yu, W., et al. (2022). Spatial transcriptomics prediction from histology jointly through Transformer and graph neural networks. Briefings in Bioinformatics, 23(5), bbac297. https://doi.org/10.1093/bib/bbac297.
- Jia, Y., Liu, J., Chen, L., et al. (2023). THItoGene: a deep learning method for predicting spatial transcriptomics from histological images. Briefings in Bioinformatics, 25. https://doi.org/10.1093/bib/bbad446.
- Xie, R., Pang, K., Chung, S., et al. (2023). Spatially resolved gene expression prediction from histology images via bi-modal contrastive learning. Advances in Neural Information Processing Systems, 36, 70626–70637.
- Min, W., Shi, Z., Zhang, J., et al. (2024). Multimodal contrastive learning for spatial gene expression prediction using histology images. Briefings in Bioinformatics, 25(6). https://doi.org/10.1093/bib/bbae469.
- Tancik, M., Srinivasan, P., Mildenhall, B., et al. (2020). Fourier features let networks learn high frequency functions in low dimensional domains. Advances in Neural Information Processing Systems, 33, 7537–7547.
- Andersson, A., Bergenstråhle, J., Asp, M., et al. (2020). Single-cell and spatial transcriptomics enables probabilistic inference of cell type topography. Communications Biology, 3(1), 565. https://doi.org/10.1038/s42003-020-01247-y.
- Ji, A. L., Rubin, A. J., Thrane, K., et al. (2020). Multimodal Analysis of Composition and Spatial Architecture in Human Squamous Cell Carcinoma. Cell, 182(2), 497–514.e22. https://doi.org/10.1016/j.cell.2020.05.032.
- Wu, S. Z., Al-Eryani, G., Roden, D. L., et al. (2021). A single-cell and spatially resolved atlas of human breast cancers. Nature Genetics, 53(9), 1334–1347. https://doi.org/10.1038/s41588-021-00911-1.
- Perez, E., Strub, F., De Vries, H., et al. (2018). FILM: Visual reasoning with a general conditioning layer. Proceedings of the AAAI Conference on Artificial Intelligence.