[ENH] Adding new checks and features to NiftiLabelsMasker reporting

[mtorabi59]

Is there an existing issue for this?

• I have searched the existing issues

Describe your proposed enhancement in detail.

Currently the NiftiLabelsMasker reporting is executed only after fit and before the data is transformed. This causes the reporting to not be aware of the changes that are going to happen for the atlas, e.g. due to masking or resampling. It would be great if the reporting was also checking and also previewing how the atlas and its regions will look after applying resampling and masking.

Benefits to the change

This change will help the user be more aware of the impact of resampling or masking they are applying to the data when applying maskers. For instance, if some regions are going to be dropped due to resampling or masking, the reporting will inform the user.

Pseudocode for the new behavior, if applicable

# insert your code below

[bthirion]

I'm unsure about that: I think that the report should represented he fitted object. When the transform method() is used, the internal state of the object no longer changes, it becomes like a function with a simple input/output logic. So I don't see the point of having report related to the transform.

[mtorabi59]

@bthirion It's not about having a report for transform of the data, but about how the atlas will look after resampling and masking. Thing is for now, as far as I understood, resampling and masking are not applied even on the atlas until transform is called. So we won't be aware of their impact on the atlas and its regions before we transform the data. We could add some functions to apply resampling and masking only on the atlas before transforming the data. However, this requires the user to pass the data if the resampling target is the data.
I don't know if overall it worths the hassle.

[bthirion]

Not sure either. This is something to discuss in a forthcoming meeting.