Was your built-in Unet model for nucleus segmentation trained with tissue or cell line? If the latter, do you have one from tissue.
The UNet model was trained on cell line images only (all images are publicly available, we used BBBC39, see https://data.broadinstitute.org/bbbc/BBBC039/ and data from https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1005177 ). We do not have a model trained on tissue images.
Many thanks for your clarification.