Missing neurodata type for static microscopy that is linked to `ImagingPlane` metadata

[h-mayorquin]

Microscopy experiments commonly produce static images (not time series) for the following use cases:

• Anatomical reference images captured before/after time series acquisition
• Field-of-view documentation showing the imaging area
• Lesion or pathology documentation
• Maximum intensity projections from functional recordings
• Registration landmarks for aligning datasets
• Z-stack acquisitions at a single timepoint

These static images have:

• Image data (2D single images or 3D z-stacks)
• The same spatial reference frame as time series data
• The same acquisition metadata (excitation wavelength, indicator, optical channels, device)
• No temporal dynamics (they are snapshots, not recordings over time)

Currently, time series types like TwoPhotonSeries cannot be used because they are designed for temporal recordings with timestamps and rates. However, there is no way to connect static microscopy images to the imaging metadata stored in ImagingPlane. The existing Image type has no mechanism to link to ImagingPlane, so users must store static images without critical spatial and acquisition metadata.

This creates a gap where:

• Static images lose their connection to microscopy metadata (excitation wavelength, indicator, anatomical location, grid spacing, optical channels, device)
• There is no way to relate static reference images to functional recordings or ROI segmentations that share the same spatial reference frame
• Complete microscopy datasets cannot be properly documented

I think we should create a new neurodata type MicroscopyImage in the nwb.ophys module for static microscopy images.

c.c. @alessandratrapani

[CodyCBakerPhD]

I will note historically that the solution has been to use the classic ...PhotonSeries with only a single frame and a rate of -1 (or a single timestamp at w/e makes sense for the image to be taken outside of normal 'session time', including NaN)

Is there any data type for this on https://github.com/catalystneuro/ndx-microscopy that you can point to? Or raise/transfer issue over there to correspond with the NWBEP where this might fit in better with respect to all the other microscopy improvements that are needed?

[h-mayorquin]

I will note historically that the solution has been to use the classic ...PhotonSeries with only a single frame and a rate of -1 (or a single timestamp at w/e makes sense for the image to be taken outside of normal 'session time', including NaN)

Yes, there are two drawbacks of this convention:

1. I think this goes against the Series semantics
2. The inspector is flagging it and confusing the users

Is there any data type for this on https://github.com/catalystneuro/ndx-microscopy that you can point to? Or raise/transfer issue over there to correspond with the NWBEP where this might fit in better with respect to all the other microscopy improvements that are needed?

Not yet. But I don't really want to increase the scope of the extension and the proposal:
catalystneuro/ndx-microscopy#7

[alessandratrapani]

I will note historically that the solution has been to use the classic ...PhotonSeries with only a single frame and a rate of -1 (or a single timestamp at w/e makes sense for the image to be taken outside of normal 'session time', including NaN)

Is there any data type for this on https://github.com/catalystneuro/ndx-microscopy that you can point to? Or raise/transfer issue over there to correspond with the NWBEP where this might fit in better with respect to all the other microscopy improvements that are needed?

Recently added: catalystneuro/ndx-microscopy#74