Summary: CSM is an R package designed as a toolbox to analyze spatially resolved tissue data. To see a demo of CSM capabilities please see the related publication.
- Requirements and installation
- Main outline of CSM
- Publication and associated datasets
- Citation
- Reporting Bugs
- CSM was developed using R version > 4.3
- CSM depends on many packages. However not all suggested packages are required to perform specific tasks. Functions will return an error if packages they depend on are not installed.
- CSM includes >100 functions organized in 6+1 modules.
- CSM provides several images and datasets to test function capabilities.
- You can install CSM package in R using the following code:
devtools::install_github("Alvaro-LJ/CSM")- You can also download CSM vignettes by setting the build_vignettes argument to TRUE:
devtools::install_github("Alvaro-LJ/CSM", build_vignettes = TRUE)
-
MODULE 0: Data generation and formatting
- Divide large images into tiles. Useful to work with whole slide images.
- Perform pixel thresholding and quantification. Aimed at scoring extra-cellular biomolecules.
- Extract color channels from RGB images. May be required to work with HE, IHC or other histochemical staining techniques.
- Perform cell segmentation and feature extraction from images.
- Arrange cell feature matrices into an adequate format.
- Arrange image metadata into an adequate format.
- Perform general quality checks and check computational resources available.
-
MODULE 1: Data normalization
- Normalize cell feature expression data to account for image and slide belonging.
-
MODULE 2: Data thresholding
- Define positive thresholds for features of interest. Thresholded data can be used to assign cell labels.
-
MODULE 3: Cell phenotype labeling
- Define cell phenotype labels based on thresholded features (see MODULE 2).
- Define cell phenotype labels based on unsupervised clustering.
- Define cell phenotype labels based on semi-supervised clustering.
- Define cell phenotype labels based on user trained algorithms.
- Check concordance between different cell phenotyping methods.
- Assign cell phenotype labels by reaching a consensus using various methods.
- Quantify cell phenotypes and find associations with image metadata.
-
MODULE 4: Heterogeneity assessment
- Calculate global cell composition heterogeneity.
- Calculate spatial heterogeneity using tiling approaches as well as texture feature analysis.
- Estimate the modifiable areal unit problem (MAUP) associated with image tiling.
-
MODULE 5: Cell-Cell spatial associations
- Calculate spatial associations between pairs and thriads of cell types.
- Calculate spatial association occurring by chance.
-
MODULE 6: Neighborhood analysis and tissue structures
- Calculate cellular neighborhoods using various algorithms.
- Analyze morphologic characteristics of discrete tissue cellular structures.
- Divide tissue into compartments according to a single cell type (for example Tumor/Stromal compartments).
To see examples of use of CSM you can have a look at CSM associated publication.
This publication has an associated GitHub repository where user can find test datasets and examples of use of CSM.
Please cite this paper in case our method or parts of it were helpful in your work.
@article{
title={Comprehensive Spatial Methods (CSM): a toolbox for spatially analyzing tissues in histopathology},
author={Alvaro Lopez-Janeiro, Eduardo Miraval-Wong, Paulo Perez-Dominguez, Raluca Alexandru, Maria Guadalupe García Vazquez, David Hardisson, Alberto Peláez-Garcia, David Ruiz-Guillamon, Ignacio Melero, Carlos E de Andrea},
journal={Laboratory Investigation},
year={2026},
doi={10.1016/j.labinv.2025.104276}
}If you encounter any bugs or unexpected behavior, please help us improve by reporting them!
You can report issues directly through the GitHub Issues page. When submitting a bug report, please include:
- A clear description of the problem
- A Min-Reprex demonstrating the abnormal behavior
- Your environment (OS, R version...)
We appreciate your feedback and contributions!