Integrative cell bin segmentation on spatial transcriptomics by Voronoi

Ming Lin

DOI: https://doi.org/10.1101/2024.11.07.622434

2024-11-08

Abstract:Spatial transcriptomics is undergoing rapid advancements and iterations. It is a beneficial tool to significantly enhance our understanding of tissue organization and relationships between cells. Recent technological advancements have achieved subcellular resolution, providing much denser spot placement for downstream analysis. A key challenge for this following analysis is accurate cell segmentation and the assignment of spots to individual cells. The primary objective of this study was to evaluate the effectiveness of a new cell segmentation approach based on subcellular level spatial transcriptomic data by confirming nuclei positions and using Voronoi diagrams, compared to direct clustering with cellbin data. Our findings demonstrate that the Voronoi method not only outperforms traditional methods in providing clearer boundaries and better separation of cell types but also excels in preserving the most transcripts, addressing the issue of low capture efficiency. This integrative methodology presents a substantial advancement in spatial transcriptomics, offering improved cell type classification and spatial pattern recognition. All codes in the paper are available at the GitHub repository: https://github.com/Charlottttttte/Cell_Segmentation_Voronoi

Genetics

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to achieve accurate cell segmentation and the assignment of spots to individual cells in spatial transcriptomics. Specifically, the authors aim to evaluate the effectiveness of a new cell - segmentation method based on Voronoi diagrams and compare it with the direct clustering method. By identifying the positions of cell nuclei and using Voronoi diagrams, this method can not only provide clearer cell boundaries and better cell - type separation, but also retain more transcript information, thus alleviating the problem of low - capture efficiency.

### Detailed Explanation

1. **Background and Challenges**
- Spatial transcriptomics technology is developing rapidly and can provide a denser distribution of spots at the sub - cellular level, which is very beneficial for downstream analysis.
- The key challenge lies in how to accurately perform cell segmentation and assign spots to individual cells.

2. **Research Objectives**
- Evaluate the effectiveness of a new cell - segmentation method based on sub - cellular - level spatial transcriptomics data.
- Compare the performance of the Voronoi method and the direct clustering method in cell segmentation.

3. **Main Contributions**
- The Voronoi method uses the positions of cell nuclei as seed points and utilizes Voronoi diagrams to divide cell regions, ensuring that each cell nucleus is assigned to a unique region.
- This method not only provides clearer cell boundaries and better cell - type separation, but also retains more transcript information, solving the problem of low - capture efficiency.

4. **Experimental Verification**
- A preliminary test was carried out on a small area (72 cells) of mouse brain tissue to verify the feasibility of the method.
- It was further verified by expanding to a larger and more complex 3000x3000 μm mouse brain tissue section, and the results showed that the Voronoi method also performed excellently in a complex tissue environment.

5. **Result Comparison**
- The Leiden clustering algorithm was used to compare the results of the original method and the Voronoi method. The clusters generated by the Voronoi method were more refined and distinct, with clearer boundaries.
- UMAP embedding analysis also indicated that the clusters generated by the Voronoi method were more dispersed and had less overlap, showing that it was more accurate in identifying cell identities.
- Quantitative analysis showed that the Silhouette Score and Calinski - Harabasz Index of the Voronoi method were significantly higher than those of the original method, further confirming its superiority.

6. **Conclusion**
- The Voronoi segmentation method shows significant improvement in cell - type classification and spatial analysis, providing a more accurate and clear spatial - hierarchical representation, which is more consistent with the known anatomical structure.
- This method is not only applicable to mouse brain tissue, but also provides new tools and methods for in - depth research on other tissues.

### Formula Presentation
Some formulas and calculation methods involved in the paper are as follows:

- **Gaussian Fit**:
\[
f(x)=A\exp\left(-\frac{(x - \mu)^2}{2\sigma^2}\right)
\]
where \(A\) is the amplitude, \(\mu\) is the mean, and \(\sigma\) is the standard deviation.

- **R² Value**:
\[
R^2 = 1-\frac{\sum(y_i-\hat{y}_i)^2}{\sum(y_i - \bar{y})^2}
\]
where \(y_i\) is the actual value, \(\hat{y}_i\) is the predicted value, and \(\bar{y}\) is the average value.

- **Chi - square Statistic**:
\[
\chi^2=\sum\frac{(O_i - E_i)^2}{E_i}
\]
where \(O_i\) is the observed value and \(E_i\) is the expected value.

Through these formulas and methods, the authors demonstrate the superior performance of the Voronoi method in spatial transcriptomics, especially the improvements in cell segmentation and gene expression analysis.