STCS: Spatial Transcriptomics Cell Segmentation

STCS (Spatial Transcriptomics Cell Segmentation) is a platform-agnostic framework that reconstructs cell-level gene expression profiles from sequencing-based spatial transcriptomics data by integrating nuclei segmentation, transcriptomic similarity, and spatial proximity.

Sequencing-based spatial transcriptomics technologies such as Visium HD and Stereo-seq provide transcriptome-wide measurements at very high spatial resolution. However, these platforms measure gene expression from spatial bins rather than biological cells, making downstream cell-level analysis challenging.

STCS addresses this problem by reconstructing coherent cell-level expression profiles through a joint transcriptomic–spatial assignment model.

---

Overview of the STCS Pipeline

The STCS pipeline consists of the following steps:

1. Nuclei Segmentation
   H&E images are processed using StarDist to detect nuclei.

2. Initial Bin Assignment
   Spatial bins located within detected nuclei are assigned to the corresponding nucleus.

3. Candidate Nucleus Search
   Bins outside nuclei search for nearby nuclei within a specified search radius (S).

4. Joint Transcriptomic–Spatial Distance Calculation

   The assignment score between bin i and nucleus c is computed based on:

   • S (search radius) defines the spatial neighborhood for candidate nuclei.

   • λ (lambda) controls the weight of spatial distance relative to transcriptomic similarity.

5. Cell Reconstruction

   Bins assigned to each nucleus are aggregated to form cell-level expression profiles.

6. Cell Type Annotation

   Reconstructed cells can be annotated using CellTypist or other cell-type annotation tools.

Visium HD Workflow

For Visium HD datasets, we recommend performing parameter tuning before running the full STCS pipeline for new tissue slides.

Step 1 — Parameter tuning

Run: Parameter_Tuning.ipynb

This notebook searches combinations of:

• Search radius (S)

• Spatial weight (λ)

and evaluates them using several metrics:

• connection score (spatial coherence)

• detected genes per cell

• cell-type stability

These metrics help identify parameters that produce stable cell reconstructions and coherent spatial structures.

Step 2 — Run the STCS pipeline

After selecting parameters, run: STCS_visium_tutorial.ipynb

Stereo-seq Workflow

Stereo-seq datasets are provided as GEM files, which must first be converted into AnnData format before running STCS.

Step 1 — Convert GEM to AnnData

Run: Convert_GEM_h5ad.ipynb

This notebook converts Stereo-seq GEM files into AnnData (.h5ad) format compatible with the STCS pipeline.

Step 2 — Run STCS on Stereo-seq data

You may run parameter tuning or run the tutorial directly: STCS_stereo-seq_tutorial.ipynb

---

Example Datasets

Visium HD Dataset

Human lung cancer dataset used in the tutorial:

Visium HD Spatial Gene Expression Libraries, Post-Xenium, Human Lung Cancer (FFPE) https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-human-lung-cancer-post-xenium-expt

Corresponding histology image: https://www.10xgenomics.com/datasets/xenium-human-lung-cancer-post-xenium-technote

Stereo-seq Dataset

Stereo-seq mouse brain dataset: https://en.stomics.tech/col1241/index.html

---

Citation

If you use STCS in your research, please cite:

Chen Wu L*, Hu X*, Zhan F, Sun C, Gonzales J, Ofer R, Tran T, Verzi MP, Liu L†, Yang J†

STCS: A Platform-Agnostic Framework for Cell-Level Reconstruction in Sequencing-Based Spatial Transcriptomics