Add on: Pre-trained CellPose models for prediction of instance-level segmentation and labeling in both 2D and 3D data

This Xtra processes a 2D image or 3D volume to generate a labeled image in which each detected object is segmented and assigned a unique numeric label. It uses various pretrained Cellpose models, each based on different architectures and training datasets tailored to specific applications. The current implementation supports only single-channel (grayscale) inputs; for models that require multiple channels, the module duplicates the grayscale image across additional channels to approximate a multi-channel input..