Anatomical regions for ROIs in ophys experiments

The ecephys visual coding dataset (with Neuropixels) gives the CCF coordinates for each unit, and I have used this to group / cluster neural responses based on their anatomical region.

I’m now gearing up to run a similar analysis on the analogous ophys dataset. For each session, I see that there is a 'targeted_structure' field in the session metadata (returned when I call data_cache.get_ophys_experiments(...)).

Is it safe to assume that every ROI in this session belongs to the anatomical region specified by this 'targeted_structure' field? Or is there more granular information that I should be using? I’m okay with there being some small amount of error in these anatomical labels, but would like to stick on the safe side if possible. In an ideal world, I’d like to get the CCF coordinates for each ROI.

Hi Alex,

We do not have CCF coordinates for each ROI for the ophys dataset, unfortunately. For the vast majority of cells the targeted structure field is accurate. The area where there is some deviation is VISrl - which is a tricky area to target. We have determined that ~25% of the ROIs that are imaged in VISrl are outside of the area boundaries determined using ISI. If you’d like that information, I can get you a list of the cell ids for those ROIs. But other than that, the targeted structure metadata is reliable.

1 Like

Thanks! I will use the targeted_structure as the ground truth label and will keep the caveats about VISrl in mind.