Skip to content
DraconiticResearch
Search Sources API
206,698 studies31 sourcesUpdated 6 min ago
Peer-reviewedCC BY-NC-ND 4.0Nature Methods

Benchmarking segmentation methods for spatial transcriptomics

  • Xabier Martínez-de-Morentin1
  • Sumeer A. Khan1,2
  • Alberto Maillo1
  • Robert Lehmann1
  • Narsis A. Kiani3
  • Vincenzo Lagani1,2,4
  • David Gomez-Cabrero1
  • Jesper Tegner1,2,5,6,7
7 affiliations
  1. 1Biomedical Sciences Division, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia
  2. 2SDAIA-KAUST Center of Excellence in Data Science and Artificial Intelligence, Thuwal, Saudi Arabia
  3. 3Algorithmic Dynamic Lab, Department of Oncology and Pathology, Center for Molecular Medicine, Karolinska Institute, Stockholm, Sweden
  4. 4Institute of Chemical Biology, Ilia State University, Tbilisi, Georgia
  5. 5Unit of Computational Medicine, Department of Medicine, Center for Molecular Medicine, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden
  6. 6Computer, Electrical and Mathematical Sciences and Engineering Division, King Abdullah University of Science and Technology (KAUST), Thuwal, Saudi Arabia
  7. 7Science for Life Laboratory, Solna, Sweden
Read full textDOICite

Abstract

Cell segmentation is a critical yet under-standardized step in spatial transcriptomics, directly affecting cell identity assignment, spatial organization and downstream biological interpretation. Despite rapid advances in spatial transcriptomics technologies, systematic evaluations of segmentation methods across platforms and biological contexts remain limited. Here we benchmark 11 image-based, transcriptomics-based and multimodal segmentation methods across 15 datasets generated by 5 spatial transcriptomics technologies, spanning 12 human and mouse tissue types. We introduce biologically informed evaluation metrics that quantify segmentation quality, transcript recovery, cell composition, morphology, expression purity and spatial organization. Benchmarking reveals substantial variation in performance across datasets, technologies and evaluation criteria, highlighting trade-offs among segmentation completeness, transcriptomic fidelity and biological interpretability. Cross-technology analyses further demonstrate that segmentation choice critically shapes downstream analyses, including cell-type specificity, transcript contamination and ligand–receptor inference. Together, our unified Nextflow pipeline provides a scalable foundation for standardized evaluation of segmentation and method selection in spatial transcriptomics.

Related

  1. Data-Adaptive Grassmann Manifold Representations for Spatial Transcriptomics Alignment
    Xiang Xiang Wang, Sean Cottrell, Guo-Wei Wei
    arXivPreprintarXiv licenceFull textAlgebra & number theoryarXiv
  2. Publications as benchmarks: auditing AI agents on spatial transcriptomics alignment
    Yiqun T. Chen, Jan Matthias, Jianing Yao +2
    arXivPreprintCC BY-NC-ND 4.0Full textGenetics & genomicsStatisticsarXiv
  3. S3DNet: Support-Driven Spatial–Semantic Decoupling for Few-Shot Strip-Steel Surface Defect Segmentation
    Ju Zhao, Jingguang Yang, Fengling Lei +4
    MathematicsPeer-reviewedCC BY 4.0General mathematicsCrossref
  4. Spatial Transcriptomics As Rasterized Image Tensors (STARIT) characterizes cell states with subcellular molecular heterogeneity
    Dee Velazquez, Caleb Hallinan, Roujin An +2
    PLOS Computational BiologyPeer-reviewedCC BYFull textBioinformatics & computational biologyGenetics & genomicsPLOSEurope PMC
  5. Modified Marrone–Treanor Dissociation Model: Formulation and Benchmarking for Diatom–Atom Mixtures
    Ross S. Chaudhry, Erik M. Torres, Thomas E. Schwartzentruber +1
    AIAA JournalPeer-reviewedLicence not statedFull text5 citedGeneral mathematicsOpenAlex

Cite

BibTeXRIS
BibTeX
@article{martinezdemorentin2026benchmarking,
  title = {{Benchmarking segmentation methods for spatial transcriptomics}},
  author = {Xabier Martínez-de-Morentin and Sumeer A. Khan and Alberto Maillo and Robert Lehmann and Narsis A. Kiani and Vincenzo Lagani and David Gomez-Cabrero and Jesper Tegner},
  journal = {Nature Methods},
  year = {2026},
  month = oct,
  doi = {10.1038/s41592-026-03272-x},
  url = {https://doi.org/10.1038/s41592-026-03272-x},
  note = {CC BY-NC-ND 4.0},
}
RIS
TY  - JOUR
TI  - Benchmarking segmentation methods for spatial transcriptomics
AU  - Xabier Martínez-de-Morentin
AU  - Sumeer A. Khan
AU  - Alberto Maillo
AU  - Robert Lehmann
AU  - Narsis A. Kiani
AU  - Vincenzo Lagani
AU  - David Gomez-Cabrero
AU  - Jesper Tegner
PY  - 2026
DA  - 2026/10/23
JO  - Nature Methods
DO  - 10.1038/s41592-026-03272-x
UR  - https://doi.org/10.1038/s41592-026-03272-x
AB  - Cell segmentation is a critical yet under-standardized step in spatial transcriptomics, directly affecting cell identity assignment, spatial organization and downstream biological interpretation. Despite rapid advances in spatial transcriptomics technologies, systematic evaluations of segmentation methods across platforms and biological contexts remain limited. Here we benchmark 11 image-based, transcriptomics-based and multimodal segmentation methods across 15 datasets generated by 5 spatial transcriptomics technologies, spanning 12 human and mouse tissue types. We introduce biologically informed evaluation metrics that quantify segmentation quality, transcript recovery, cell composition, morphology, expression purity and spatial organization. Benchmarking reveals substantial variation in performance across datasets, technologies and evaluation criteria, highlighting trade-offs among segmentation completeness, transcriptomic fidelity and biological interpretability. Cross-technology analyses further demonstrate that segmentation choice critically shapes downstream analyses, including cell-type specificity, transcript contamination and ligand–receptor inference. Together, our unified Nextflow pipeline provides a scalable foundation for standardized evaluation of segmentation and method selection in spatial transcriptomics.
SN  - 1548-7091
SN  - 1548-7105
ER  -