# SToFM: a Multi-scale Foundation Model for Spatial Transcriptomics

Source: [SToFM: a Multi-scale Foundation Model for Spatial Transcriptomics](<https://www.sota2.com/research/paper/68fa5f7c1e209131ce9ea3cc>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-25T19:18:45.922Z

Authors: Suyuan Zhao, Yizhen Luo, Ganbo Yang, Yan Zhong, Hao Zhou, Zaiqing Nie

Published: 2025-07-15T14:47:01.000Z

## Abstract

Spatial Transcriptomics (ST) technologies provide biologists with rich insights into single-cell biology by preserving spatial context of cells. Building foundational models for ST can significantly enhance the analysis of vast and complex data sources, unlocking new perspectives on the intricacies of biological tissues. However, modeling ST data is inherently challenging due to the need to extract multi-scale information from tissue slices containing vast numbers of cells. This process requires integrating macro-scale tissue morphology, micro-scale cellular microenvironment, and gene-scale gene expression profile. To address this challenge, we propose SToFM, a multi-scale Spatial Transcriptomics Foundation Model. SToFM first performs multi-scale information extraction on each ST slice, to construct a set of ST sub-slices that aggregate macro-, micro- and gene-scale information. Then an SE(2) Transformer is used to obtain high-quality cell representations from the sub-slices. Additionally, we construct \\textbf{SToCorpus-88M}, the largest high-resolution spatial transcriptomics corpus for pretraining. SToFM achieves outstanding performance on a variety of downstream tasks, such as tissue region semantic segmentation and cell type annotation, demonstrating its comprehensive understanding of ST data through capturing and integrating multi-scale information.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Spatial domain detection on Ton](<https://www.sota2.com/research/sota/spatial-domain-detection-on-ton.md>): Spatial domain detection; Ton; Accuracy: 65.8.

- [Spatial domain detection on Lymph Node (LNA) A1](<https://www.sota2.com/research/sota/spatial-domain-detection-on-lymph-node-lna-a1.md>): Spatial domain detection; Lymph Node (LNA) A1; Accuracy: 60.2.

- [Spatial domain detection on Col](<https://www.sota2.com/research/sota/spatial-domain-detection-on-col.md>): Spatial domain detection; Col; Accuracy: 46.1.

- [Spatial Domain Detection on DLPFC](<https://www.sota2.com/research/sota/spatial-domain-detection-on-dlpfc.md>): Spatial domain detection; DLPFC; Accuracy: 64.5.

- [Spatial domain detection on Lymph Node (LND) D1](<https://www.sota2.com/research/sota/spatial-domain-detection-on-lymph-node-lnd-d1.md>): Spatial domain detection; Lymph Node (LND) D1; Accuracy: 55.5.

- [Spatial domain detection on BrC](<https://www.sota2.com/research/sota/spatial-domain-detection-on-brc.md>): Spatial domain detection; BrC; Accuracy: 44.3.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2507.11588>)

[PDF](<https://arxiv.org/pdf/2507.11588.pdf>)
