search

JohannesWiesner / nict

Network Control Theory for Neuroimaging Data
MIT License
1 stars 0 forks source link

The upcoming TransformerClass should be able to directly mask images #1

Open JohannesWiesner opened 2 years ago

JohannesWiesner commented 2 years ago

https://github.com/JohannesWiesner/nict/blob/56c186fcf5e5280f3ebce5540a13cee1aae147d8/nict/multi_subject.py#L87

As mentioned already as comments within the code, it would be better to stick to nilearn/sciki-learn API. The whole Class should also be able to directly inherit from NiftiLabelsmasker. Then the user could directly pass 4D nifti images (where each 3D image represents one state image). Also I think flattening the adjacency matrices as it is right now leads to very slow pandas data frames.

JohannesWiesner commented 2 years ago

Note however, that this would make the class very explicit towards Nifti-Files. Some other people may have other data (e.g. non-neuroimaging) data and for these people all functions should still work. But as far as I know, all the Masker classes also accept numpy arrays? Check the docs!