Machine Learning¶
Machine learning models open up powerful new possibilities in image processing and other areas. Such models can be integrated, trained and used in XamFlow.
Machine Learning Models: Segment Anything¶
This model can be used for semi-automatic / semi-interactive segmentation of 2D structures.
Machine Learning Models: nnUNet¶
This model can be trained for automatic segmentation of 3D structures.
See:
Example.nnunet.InstallDatasetExample.nnunet2.ImportModelExample.nnunet2.ExportModelExample.nnunet2.TrainExample.nnunet2.SegmentImageWorkflow.Bone.nnunetSegmentationIPLEvaluation
Hierarchical labels are also supported.
When a model is already available, it can be imported by selecting the model file. New model files can be created by first installing a training dataset from another workflow, starting the training task, and then exporting the model file. The model only has to be imported once. The name of the model can then be configured in the inference task configuration to use it to segment images.
Machine Learning Models: MaskRCNN¶
This model can be trained for automatic instance segmentation of 2D structures.
See Example.MaskRCNN.Train and Example.MaskRCNN.SegmentImage2D.
Machine Learning Models: TotalSegmentator¶
This model can be used for automatic segmentation of anatomical structures in CT images.
See Example.TotalSegmentator.SegmentImage, Workflow.Example.TotalSegmentator and Workflow.Example.TotalSegmentatorQuantitative.
Machine Learning Models: cellpose¶
This model can be used for automatic segmentation of cells in microscopy images.
See Example.cellpose.SegmentImage.
Machine Learning Models: stardist¶
This model can be used for automatic segmentation of cells in microscopy images.
See Example.stardist.Fluorescence2D.
