Img2ST-Net: efficient high-resolution spatial omics prediction from whole-slide histology images via fully...

Recent advances in multimodal artificial intelligence (AI) have demonstrated promising potential for generating the currently expensive spatial transcriptomics (ST) data directly from routine histology images, offering a means to reduce the high cost and time-intensive nature of ST data acquisition. However, the increasing resolution of ST-particularly with platforms such as Visium HD achieving $8\mu m$ or finer-introduces significant computational and modeling challenges. Conventional spot-by-spot sequential regression frameworks become inefficient and unstable at this scale, whereas the inherent extreme sparsity and low expression levels of high-resolution ST further complicate both prediction and evaluation.

To address these limitations, we propose Img2ST-Net, a high-definition (HD) histology-to-ST generation framework for efficient and parallel high-resolution ST prediction. Unlike conventional spot-by-spot inference methods, Img2ST-Net employs a fully convolutional architecture to generate dense, HD gene expression maps in a parallelized manner. By modeling HD ST data as super-pixel representations, the task is reformulated from image-to-omics inference into a super-content image generation problem with hundreds or thousands of output channels. This design not only improves computational efficiency but also better preserves the spatial organization intrinsic to spatial omics data. To enhance robustness under sparse expression patterns, we further introduce SSIM-ST, a structural-similarity-based evaluation metric tailored for high-resolution ST analysis.

Evaluations on two public Visium HD datasets at 8 and $16\mu m$ resolutions demonstrate that Img2ST-Net outperforms state-of-the-art methods in both accuracy and spatial coherence. On the Breast Cancer dataset at $16\mu m$ , Img2ST-Net achieves a mean squared error (MSE) of 0.1657 and a structural similarity index of 0.0937, whereas on the Colorectal Cancer dataset, it reaches an MSE of 0.7981 and a mean absolute error of 0.5208. These results highlight its ability to capture fine-grained gene expression patterns. In addition, our region-wise modeling significantly reduces training time without sacrificing performance, achieving up to 28-fold acceleration over conventional spot-wise methods. Ablation studies further validate the contribution of contrastive learning in enhancing spatial fidelity. The source code has been made publicly available at https://github.com/hrlblab/Img2ST-Net.

We present a scalable, biologically coherent framework for high-resolution ST prediction. Img2ST-Net offers a principled solution for efficient and accurate ST inference at scale. Our contributions lay the groundwork for next-generation ST modeling that is robust and resolution-aware.