Skip to content

Segmentation quality assessment #7

Description

@rafalab

Segmentation algorithms in spatial transcriptomics imaging play define the boundaries of cells or regions from which gene expression is quantified. Errors or variability in segmentation, such as over-segmentation, under-segmentation, or misassignment of transcripts, can substantially distort downstream analyses, including cell type identification, spatial pattern detection, and many others. Despite this impact, AFAIK, there is currently no widely accepted standard for assessing segmentation quality, making it difficult to compare methods or ensure reproducibility across studies. I see two key challenges: 1) developing robust quality metrics based on downstream data to evaluate segmentation performance, and 2) improving the segmentation algorithms themselves to better capture true biological structure. This gap suggests a valuable opportunity to explore these issues through systematic data exploration and benchmarking during a hackathon. Tackling this challenge can also inform what aspects of class definitions and interpretability with software does the segmentation (often in Python)

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    Status
    Todo

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions