pith. sign in

arxiv: 2606.09675 · v1 · pith:5BRSIFGNnew · submitted 2026-06-08 · 🧬 q-bio.OT

The Challenge of Cell Segmentation in Spatially Resolved Transcriptomics

Naveed Ishaque , Peter Kharchenko , Daria Lazic , Jieran Sun , Jean Yee Hwa Yang , Martin Emons , Florian Heyl , Wolfgang Huber
show 15 more authors
Daniel Jones Louis B. Kuemmerle Alex R. Lederer Malte D. Luecken Vinicius Maracaja-Coutinho Matthias Meyer-Bender Andrew Moorman Evan W. Newell Quan Nguyen Shyam Prabhakar John Randell Daria Romanovskaia Oliver Stegle Gary D. Bader Raphael Gottardo
This is my paper

Pith reviewed 2026-06-27 14:05 UTC · model grok-4.3

classification 🧬 q-bio.OT
keywords cell segmentationspatially resolved transcriptomicsspatial omicstranscript assignmentbenchmarkingerror propagationtissue morphologyreproducibility
0
0 comments X

The pith

Cell segmentation should be treated as a central unresolved problem in spatial transcriptomics rather than routine preprocessing.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper argues that cell segmentation in spatially resolved transcriptomics is not a routine preprocessing task but a major source of uncertainty. Challenges such as sparse molecular signals, transcript displacement, complex cellular morphologies, and 3D-to-2D projection make accurate transcript-to-cell assignment difficult. These segmentation errors can propagate through downstream analyses and lead to misleading biological interpretations. The authors review existing approaches, highlight the lack of suitable metrics and gold-standard benchmarks, and call for community efforts to establish shared evaluation frameworks, scalable benchmark datasets, and transparent reporting standards.

Core claim

Segmentation should be treated as a central unresolved problem in spatial omics. The major technical challenges of sparse molecular signals, transcript displacement, complex cellular morphologies, and the projection of three-dimensional tissue architecture onto two-dimensional imaging planes create uncertainty. Errors from these challenges propagate through downstream analyses and can yield misleading biological interpretations. Current methods lack appropriate metrics and gold-standard benchmarks.

What carries the argument

The propagation of segmentation errors from technical challenges including sparse molecular signals, transcript displacement, complex cellular morphologies, and 3D-to-2D projection in SRT data to downstream biological analyses.

If this is right

  • Segmentation errors can propagate through downstream analyses to produce misleading biological interpretations.
  • The lack of appropriate metrics and gold-standard benchmarks limits reliable evaluation of segmentation methods.
  • Community-driven evaluation frameworks are required to advance segmentation approaches.
  • Scalable benchmark datasets and transparent reporting standards will be essential for making SRT robust and reproducible.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Better segmentation could improve the accuracy of mapping cell types and states within intact tissues.
  • The problem likely affects reproducibility across independent SRT studies using different segmentation pipelines.
  • Focused community benchmarks could speed development of algorithms that directly address the listed technical challenges.
  • Standardized segmentation practices may increase the suitability of SRT data for clinical translation.

Load-bearing premise

The listed technical challenges produce segmentation errors that propagate through downstream analyses to yield misleading biological interpretations.

What would settle it

A controlled comparison showing that different segmentation methods applied to identical SRT datasets produce consistent downstream biological interpretations would undermine the central claim.

read the original abstract

Spatially resolved transcriptomics (SRT) is transforming how we study tissues by measuring gene expression in cells in their spatial context. However, the field lacks robust methodological guidance on one of its most fundamental analytical steps: how to accurately segment cells and assign spatially localized transcripts to them. Major technical challenges include sparse molecular signals, transcript displacement, complex cellular morphologies, and the projection of three-dimensional tissue architecture onto two-dimensional imaging planes. These challenges make segmentation a major source of uncertainty, with errors that can propagate through downstream analyses and ultimately lead to misleading biological interpretations. Here, we argue that segmentation should be treated as a central unresolved problem in spatial omics rather than a routine preprocessing step. We review current approaches, highlight key methodological limitations, including the lack of appropriate metrics and gold-standard benchmarks, and propose a community-driven path forward. Establishing shared evaluation frameworks, scalable benchmark datasets, and transparent reporting standards will be essential for transforming SRT into a robust and reproducible foundation for biological discovery and clinical translation.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The manuscript is a perspective article arguing that cell segmentation should be treated as a central unresolved problem in spatially resolved transcriptomics (SRT) rather than routine preprocessing. It identifies key technical challenges (sparse molecular signals, transcript displacement, complex cellular morphologies, and 3D-to-2D projection) that create uncertainty, asserts that resulting errors can propagate through downstream analyses to produce misleading biological interpretations, reviews existing approaches and their limitations (particularly the absence of suitable metrics and gold-standard benchmarks), and proposes community-driven development of evaluation frameworks, scalable benchmark datasets, and transparent reporting standards.

Significance. If the perspective's core argument holds, it could usefully redirect attention in the SRT field toward methodological foundations, encouraging the development of shared benchmarks that would enhance reproducibility and reliability across studies. The explicit call for community standards and datasets is a constructive contribution aligned with reproducibility efforts in computational biology.

major comments (1)
  1. [Abstract, paragraph 2] Abstract, paragraph 2: The assertion that segmentation errors 'can propagate through downstream analyses and ultimately lead to misleading biological interpretations' is presented as motivation without any concrete examples, quantitative illustrations, or citations to studies demonstrating such propagation. This claim is load-bearing for the recommendation to elevate segmentation beyond routine preprocessing.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive review and positive assessment of the perspective. We address the single major comment below.

read point-by-point responses

  1. Referee: [Abstract, paragraph 2] Abstract, paragraph 2: The assertion that segmentation errors 'can propagate through downstream analyses and ultimately lead to misleading biological interpretations' is presented as motivation without any concrete examples, quantitative illustrations, or citations to studies demonstrating such propagation. This claim is load-bearing for the recommendation to elevate segmentation beyond routine preprocessing.

    Authors: We agree that the abstract states this claim without accompanying examples or citations. While the perspective develops the underlying technical challenges in detail in the main text, the referee is correct that explicit support for propagation effects would strengthen the motivation. In the revised version we will add brief, cited examples of how segmentation inaccuracies have been shown to affect downstream biological conclusions (e.g., in cell-type composition or spatial neighborhood analyses) either in the introduction or a short new subsection, while keeping the abstract concise. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The manuscript is a qualitative perspective article with no equations, derivations, fitted parameters, or quantitative predictions. Its central claim—that segmentation should be treated as an unresolved problem warranting community benchmarks—is advanced through synthesis of acknowledged technical limitations rather than any self-referential reduction or load-bearing self-citation chain. The argument rests on external literature and community needs, remaining self-contained without circular steps.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No quantitative model or derivation is present; the argument relies on domain consensus about technical difficulties rather than new postulates.

pith-pipeline@v0.9.1-grok · 5808 in / 1011 out tokens · 14478 ms · 2026-06-27T14:05:45.113127+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

18 extracted references · 12 canonical work pages

  1. [1]

    Anacleto, A. et al. (2026) ‘Seq-Scope-eXpanded: spatial omics beyond optical resolution’, Nature Communications, pp. 1–20. Andersson, A. et al. (2024) ‘Points2Regions: Fast, interactive clustering of imaging-based spatial transcriptomics data’, Cytometry. Part A: The Journal of the International Society for Analytical Cytology, 105(9), pp. 677–687. Archit...

    2026

  2. [2]

    Benjamin, K. et al. (2024) ‘Multiscale topology classifies cells in subcellular spatial transcriptomics’, Nature, 630(8018), pp. 943–949. Bilous, M. et al. (2026) ‘Resolving sensitivity, specificity and signal contamination in Xenium spatial transcriptomics’, Nature Methods, pp. 1–11. Cajigas, I.J. et al. (2012) ‘The local transcriptome in the synaptic ne...

    work page doi:10.1101/181677v1 2024

  3. [3]

    Dessimoz, C. et al. (2017). Lessons Learned: Recommendations for Establishing Critical Periodic Scientific Benchmarking. bioRxiv (Cold Spring Harbor Laboratory). doi:https://doi.org/10.1101/181677. Ding, S. et al. (2026) ‘scGPT: end-to-end protocol for fine-tuned retinal cell type annotation’, Nature protocols, 21(3), pp. 873–893. Ding, T. et al. (2025) ‘...

    work page doi:10.1101/181677 2017

  4. [4]

    Gong, D. et al. (2024) ‘Spatial oncology: Translating contextual biology to the clinic’, Cancer Cell, 42(10), pp. 1653–1675. Gortana, L. et al. (2026) ‘HEDeST: An integrative approach to enhance spatial transcriptomic deconvolution with histology’, bioRxiv. bioRxiv. Available at: https://doi.org/10.64898/2026.01.06.697922. He, H.-F. et al. (2026) ‘Unlocki...

    work page doi:10.64898/2026.01.06.697922 2024

  5. [5]

    and Regev, A

    Jerby-Arnon, L. and Regev, A. (2022) ‘DIALOGUE maps multicellular programs in tissue from single-cell or spatial transcriptomics data’, Nature biotechnology, 40(10), pp. 1467–1477. Jones, D.C. et al. (2025) ‘Cell simulation as cell segmentation’, Nature methods, 22(6), pp. 1331–

    2022

  6. [6]

    Kirillov, A. et al. (2023) ‘Segment Anything’, in 2023 IEEE/CVF International Conference on Computer Vision (ICCV). 2023 IEEE/CVF International Conference on Computer Vision (ICCV), IEEE, pp. 3992–4003. Kuehl, M. et al. (2025) ‘Pathology-oriented multiplexing enables integrative disease mapping’, Nature, 644(8076), pp. 516–526. Kwok, A.W.C. et al. (2025) ...

    work page doi:10.1101/2025.11.13.688387 2023

  7. [7]

    Littman, R. et al. (2020) ‘JSTA: joint cell segmentation and cell type annotation for spatial transcriptomics’, bioRxiv. Available at: https://doi.org/10.1101/2020.09.18.304147. Liu, C.C. et al. (2023) ‘Robust phenotyping of highly multiplexed tissue imaging data using pixel- level clustering’, Nature Communications, 14(1), p

    work page doi:10.1101/2020.09.18.304147 2020

  8. [8]

    and Wang, L

    Liu, Y., Dai, Y. and Wang, L. (2026) ‘Spatial omics at the forefront: emerging technologies, analytical innovations, and clinical applications’, Cancer Cell, 44(1), pp. 24–49. Luecken, M.D. et al. (2025) ‘Defining and benchmarking open problems in single-cell analysis’, Nature Biotechnology, 43(7), pp. 1035–1040. Mah, C.K. et al. (2024) ‘Bento: a toolkit ...

    2026

  9. [9]

    Ma, J. et al. (2024) ‘The multimodality cell segmentation challenge: toward universal solutions’, Nature Methods, 21(6), pp. 1103–1113. Mallona, I. et al. (2024) ‘Omnibenchmark: transparent, reproducible, extensible and standardized orchestration of solo and collaborative benchmarks’, arXiv [q-bio.OT]. Available at: https://doi.org/10.48550/arXiv.2409.170...

    work page doi:10.48550/arxiv.2409.17038 2024

  10. [10]

    (2012) ‘Cell segmentation: 50 years down the road [life sciences]’, IEEE Signal Processing Magazine, 29(5), pp

    Meijering, E. (2012) ‘Cell segmentation: 50 years down the road [life sciences]’, IEEE Signal Processing Magazine, 29(5), pp. 140–145. MERSCOPE TM User Guide Protein Stain Verification Kit Vizgen TM Materials*. (n.d.). Available at: https://vizgen.com/wp-content/uploads/2022/12/91600103_MERSCOPE-Protein-Stain-Verification- Kit-User-Guide_Rev-A.pdf [Access...

    work page doi:10.1101/2025.01.02.631135 2012

  11. [11]

    Available at: https://cdn.10xgenomics.com/image/upload/v1754601291/support- documents/CG000750_XeniumInSitu_CellSegmentation_TechNote_RevB.pdf (Accessed: 6 March 2026)

    [No title] (no date). Available at: https://cdn.10xgenomics.com/image/upload/v1754601291/support- documents/CG000750_XeniumInSitu_CellSegmentation_TechNote_RevB.pdf (Accessed: 6 March 2026). Oliveira, M.F. de et al. (2025) ‘High-definition spatial transcriptomic profiling of immune cell populations in colorectal cancer’, Nature genetics, 57(6), pp. 1512–1...

    2026

  12. [12]

    and Stringer, C

    Pachitariu, M., Rariden, M. and Stringer, C. (2025) ‘Cellpose-SAM: superhuman generalization for cellular segmentation’, bioRxiv. Available at: https://doi.org/10.1101/2025.04.28.651001. Pachitariu, M. and Stringer, C. (2022) ‘Cellpose 2.0: how to train your own model’, Nature Methods, 19(12), pp. 1634–1641. Park, J. et al. (2021) ‘Cell segmentation-free ...

    work page doi:10.1101/2025.04.28.651001 2025

  13. [13]

    and Wählby, C

    Partel, G. and Wählby, C. (2021) ‘Spage2vec: Unsupervised representation of localized spatial gene expression signatures’, The FEBS Journal, 288(6), pp. 1859–1870. Peters Couto, B.Z. et al. (2023) ‘MoleculeExperiment enables consistent infrastructure for molecule- resolved spatial omics data in bioconductor’, Bioinformatics (Oxford, England), 39(9), p. bt...

    2021

  14. [14]

    Qian, X. et al. (2020) ‘Probabilistic cell typing enables fine mapping of closely related cell types in situ’, Nature Methods, 17(1), pp. 101–106. Regev, A. et al. (2017) ‘The Human Cell Atlas’, eLife,

    2020

  15. [15]

    Available at: https://doi.org/10.7554/eLife.27041. Ren, Y. et al. (2024) ‘BEACON: Benchmark for comprehensive RNA tasks and Language Models’, arXiv [q-bio.QM]. Available at: https://doi.org/10.48550/arXiv.2406.10391. Resolve BioSciences. (2025). Technology | Resolve BioSciences. [online] Available at: https://www.resolvebiosciences.com/technology. Rojas-H...

    work page doi:10.7554/elife.27041 2024

  16. [16]

    Cham: Springer International Publishing (Lecture Notes in Computer Science), pp. 265–273. Schott, M. et al. (2024) ‘Open-ST: High-resolution spatial transcriptomics in 3D’, Cell, 187(15), pp. 3953–3972.e26. Seydel, C. (2025) ‘Beyond cell atlases: spatial biology reveals mechanisms behind disease’, Nature Biotechnology, 43(6), pp. 841–844. Shaban, M. et al...

    work page doi:10.3390/cancers14122916 2024

  17. [17]

    Sun, E.D. et al. (2025) ‘Spatial transcriptomic clocks reveal cell proximity effects in brain ageing’, Nature, 638(8049), pp. 160–171. Theodoris, C.V. et al. (2023) ‘Transfer learning enables predictions in network biology’, Nature, 618(7965), pp. 616–624. Tiesmeyer, S. et al. (2026) ‘Identifying 3D signal overlaps in spatial transcriptomics data with ovr...

    work page doi:10.48550/arxiv.2511.05708 2025

  18. [18]

    and Danaher, P

    Wu, L., Beechem, J.M. and Danaher, P. (2024) ‘FastReseg: using transcript locations to refine image-based cell segmentation results in spatial transcriptomics’, bioRxiv. Available at: https://doi.org/10.1101/2024.12.05.627051. Wu, L., Beechem, J.M. and Danaher, P. (2025) ‘Using transcripts to refine image based cell segmentation with FastReseg’, Scientifi...

    work page doi:10.1101/2024.12.05.627051 2024