cell-based phenotypic screening dataset data-split method

[Sixseven-Five]

Is the segmentation method used when running the model on cell-based phenotypic screening dataset for comparison scaffold or random？