Segmentations

[Erasdna]

Add Segmentation class with FreeSurfer labels, make io go through MRIData, rework of stats dataframe function.

Tests are passing.

[finsberg]

Thanks @Erasdna for the PR! I fixed the merge conflicts and made some changes. Would be great if you could have a look an see if it make sense? I am also not sure about the changes in the compute_stats module. Would be good to get an opinion from @cdaversin as well.

@cdaversin could you take a look?

[Erasdna]

Thanks!

Part of my thinking was that it could be useful to be able to specify which statistics you want to compute, and then store them as a pivot table, i.e, one value column and multiple grouping columns (which I think is often an advantage for later processing). Maybe the two generate dataframe functions should operate in a more similar way? Computing an unnecessary number of statistics can sometimes take too much time imo (but the specific implementation of this is maybe not optimal as it stands)

I also think it would be useful to decide on whether functions should take a Path or an MRIdata/Segmentation object as input (this applies several places). I would like to add some more functionality to modify segmentations etc, which would make passing the class object a bit more practical, I think.

I am also curious about the regex stuff @cdaversin, to me it seems a little bit convoluted sometimes. I added a way to directly read patient info assuming that files follow the bids convention (which is sota I believe). Would maybe be nice to stick to that? We should probably also make some way to generate filenames according to bids?