CellTune: human-in-the-loop classification for spatial proteomics

Authors: Yuval Bussi, Dana Shainshein, Leeat Keren and collaborators. Venue: Nature Methods, peer reviewed. Contribution: active learning combines analyst feedback with cell classification for spatial proteomics. Significance: a practical example of human approval improving specialised ML pipelines. Limitations: domain-specific; performance depends on data quality and expert annotations. Nature listing.