The Multimodality Cell Segmentation Challenge: Toward Universal Solutions

Cell segmentation is a critical step for quantitative single-cell analysis inmicroscopy images. Existing cell segmentation methods are often tailored tospecific modalities or require manual interventions to specify hyper-parametersin different experimental settings. Here, we present a multi-modality cellsegmentation benchmark, comprising over 1500 labeled images derived from morethan 50 diverse biological experiments. The top participants developed aTransformer-based deep-learning algorithm that not only exceeds existingmethods but can also be applied to diverse microscopy images across imagingplatforms and tissue types without manual parameter adjustments. This benchmarkand the improved algorithm offer promising avenues for more accurate andversatile cell analysis in microscopy imaging.