PhenoMe: An interactive AI framework for interpretable, microscopy-driven phenotyping

PhenoMe is an open-source, model-agnostic framework that enables interpretable, microscopy-driven phenotyping across seven public microscopy datasets using foundation-model image embeddings.

• Why it matters: Microscopy datasets need quantitative tools that connect visual similarity with experimental context, but existing workflows often lack interpretable links between embeddings and measurable image properties.
• What they did: PhenoMe combines pretrained foundation-model-derived embeddings with quantitative image properties and experimental metadata, offering interactive embedding visualization, dimensionality reduction, clustering, similarity retrieval, and correlation analyses.
• The result: Case studies across seven datasets show that readily available representations can support contextualisation, curation, and analysis, making representation-based microscopy exploration accessible via notebooks, reproducible configs, and a napari plugin.

AI summary of the abstract · may contain errors · not medical advice · read the paper.