apply_hilbert for source estimated data

[BabaSanfour]

Describe the new feature or enhancement

Currently apply_hilbert can be applied only to raw/epochs/epoched data through, for example, epochs. apply_hilbert. The idea is to change this whole function into the time_frequency module. Then the epochs/raw/evoked/source data can be an argument to the function (EX: mne.time_frequency.EpochsTFR)

Describe your proposed implementation

It can be a new class in the tfr.py file. It would be based on the ancient .apply_hilbert function.

Describe possible alternatives

The alternative that is being used now is script hilbert function for source data or apply_hilbert for other types. The proposed implementation will make the function independent of the data type/object

Additional context

I am happy to propose an implementation. I just want to check with maintainers for directions before starting. As this is a huge change in a lot of classes/files. I think it should be discussed before starting the implementation.

[welcome[bot]]

Hello! 👋 Thanks for opening your first issue here! ❤️ We will try to get back to you soon. 🚴

[larsoner]

Alternatively we could make FilterMixin work for SourceEstimate instances. I think in principle it should be doable -- would need to make FilterMixin.filter, FilterMixin.savgol_filter, and FilterMixin.resample (which already exists for SourceEstimate classes) work properly. WDYT?

[BabaSanfour]

Yes that would work too; By "already exists for SourceEstimate classes" you refer to the resample function only right? could not find the other implementations.

[larsoner]

Yeah just resample exists currently

[BabaSanfour]

ok thanks! I will take that into account when making the changes.

[BabaSanfour]

@larsoner the function has been added