SegmentData[data]
segments the data using the default "Body" method.
SegmentData[data,what]
segments using the specified anatomical region. With full body output labels what can be "Body", "LegsBody", "LegsHipBody", "HipBody", "UpperLegsBody", "LowerLegsBody", or "ShoulderBody". With legacy output labels what can be "Legs", "LegsHip", "UpperLegs", "LowerLegs", "Hip", "Shoulder", or "Arm".
SegmentData[{data,vox}]
and
SegmentData[{data,vox},what]
additionally give the voxel size of data. Unless vox is {1, 1, 1}, data is rescaled before segmentation to {vox[[1]], 1.5, 1.5} for SegmentationDimension->"2D" or to {6, 1.5, 1.5} for "3D". If NetworkOutput->"Both" returns {segmentation, confidence}, any other value always returns just the segmentation.
SegmentData
SegmentData[data]
segments the data using the default "Body" method.
SegmentData[data,what]
segments using the specified anatomical region. With full body output labels what can be "Body", "LegsBody", "LegsHipBody", "HipBody", "UpperLegsBody", "LowerLegsBody", or "ShoulderBody". With legacy output labels what can be "Legs", "LegsHip", "UpperLegs", "LowerLegs", "Hip", "Shoulder", or "Arm".
SegmentData[{data,vox}]
and
SegmentData[{data,vox},what]
additionally give the voxel size of data. Unless vox is {1, 1, 1}, data is rescaled before segmentation to {vox[[1]], 1.5, 1.5} for SegmentationDimension->"2D" or to {6, 1.5, 1.5} for "3D". If NetworkOutput->"Both" returns {segmentation, confidence}, any other value always returns just the segmentation.
Details and Options
- The following options can be given:
-
MaxMemorySize Automatic MaxMemorySize is an option for SegmentData and ApplySegmentationNetwork. Defines the maximum memory a patch can use in Gb. Higher is better. SegmentationDimension "3D" SegmentationDimension is an option for SegmentData. The default value is "3D" but can also be set to "2D" to used 2D slice by slice segmentation. NetworkOutput "Segmentation" NetworkOutput is an option for ApplySegmentationNetwork. Can be "Segmentation" (default) to return the segmentation, "Confidence" to return the per-voxel class confidence (see ClassConfidence), "Both" to return {segmentation, confidence}, or "Volume" for a network that outputs a reconstructed volume rather than class probabilities (e.g. a self-supervised pretrained network), which merges patches without ClassDecoder. Monitor False Monitor[expr,mon] generates a temporary monitor cell in which the continually updated current value of mon is displayed during the course of evaluation of expr.