Oct 2026· Scientific Reports· Vol 16· 0 citations· 15 references
Medicine
TL;DR
This work describes a benchmarking approach for rapid assessment of segmentation algorithm performance in melanoma and breast cancer contexts using mass spectrometer and optical spatial omics imaging platforms and identified Cellpose models that performed well across different tumor samples and data sources with minimal fine-tuning.
Abstract
The in situ tissue microenvironment is a complex mixture of different cell types, microanatomic structures, and acellular matrix material. Further, disease perturbs the morphologies of these features significantly. Spatial proteomics utilizing optical or mass spectrometer imaging simultaneously detects expression and subcellular localization of many proteins. Accurate identification of heterogeneous, irregular, and tightly interdigitating cell boundaries is a crucial step for all downstream analyses, including cell phenotyping and cell-cell interaction networks. We describe a benchmarking approach for rapid assessment of segmentation algorithm performance in melanoma and breast cancer contexts using mass spectrometer and optical spatial omics imaging platforms. Comparison of cell segmentation by fourteen methods to pathologist ground-truth annotation with Jaccard pixel index and F1 object measurement identified Cellpose models that performed well across different tumor samples and data sources with minimal fine-tuning. Algorithms utilizing cell membrane and nuclear features demonstrated better performance in both tumor samples. This benchmarking approach is a rapid and effective method to assess performance of cell segmentation methods in different contexts for efficient algorithm selection and iterative model training.
In spatial transcriptomics, cell segmentation is used to draw boundaries around cells. Molecules located within a cell’s boundary are assigned to it, making its gene expression profile dependent on segmentation accuracy. To identify potentially problematic cells, studies increasingly use doublet detection methods, a cl...
Tissue imaging depends on cell segmentation, whose errors carry into downstream analyses. In human brain, pretrained models struggle with complex cell bodies, and the segmentation accuracy has rarely been evaluated. Here, we introduce SPInfer, a workflow developed using thousands of expert-annotated cell bodies from 12...
Liang You, Jamie L. Durbin, Guan-Qun Meng et al.· bioRxiv· 0 citations
To benchmark multiple segmentation approaches in cardiac spatial transcriptomics using post-acquisition membrane staining as biologically defined ground truth, and to determine whether combining image-based and transcript-based methods improves cardiomyocyte recovery, seven segmentation algorithms are being benchmarked...
H. Tao, A. Lewis· European Heart Journal· 0 citations
It is argued that cell segmentation for spatial pathology should be evaluated not only by boundary and instance accuracy, but also by preservation of the biological relationships that downstream analyses are intended to measure.
Nevenka Dimitrova, Marjan Stoimchev, A. Filliol et al.· bioRxiv· 0 citations
A multimodal pseudo-ground truth training technique to enable supervised feature learning on H&E images for whole-cell segmentation and adapted into a ST pipeline to create an integrated end-to-end machine learning solution, without additional optimisation on ST data.
A.-A. Zhang, Ulrich D. Kadolsky, A. Saxena· 0 citations