We already allow a Volume imaging mode, and the acquisitions can be 3D instead of 2D with small changes. However, Plane segmentations are not a good fit for 3D ROIs.
We could shoehorn them into this scheme (e.g. store each ROI under the plane of its starting z coordinate), but ideally we can define an extension for 3D segmentations and use that instead.
We already allow a
Volumeimaging mode, and the acquisitions can be 3D instead of 2D with small changes. However, Plane segmentations are not a good fit for 3D ROIs. We could shoehorn them into this scheme (e.g. store each ROI under the plane of its starting z coordinate), but ideally we can define an extension for 3D segmentations and use that instead.Can possibly try with the data from 2016 Nature Methods paper?
EDIT: Spoke too soon! The
PlaneSegmentationshould support 3D ROIs since NeurodataWithoutBorders/pynwb#688.