mask_filepath overwritten with fmriprep pipeline when defining Fmri event type

[MarieStLaurent]

Description

I attempted to instantiate a Study class to process an fmriprep dataset (4d BOLD volumes). At the moment, the mask_filepath param becomes overwritten even when I specify it manually when instantiating Fmri event types:

fmri_row: dict[str, tp.Any] = {
"type": "Fmri",
"start": 0.0,
"duration": float(n_TRs) * tr_s,
"frequency": 1.0 / tr_s,
"filepath": bold_path,
"mask_filepath": bold_path.replace("preproc_bold.", "brain_mask."),
"confounds_filepath": bold_path.replace(
f"_space-{self.space}_desc-preproc_bold.nii.gz",
"_desc-confounds_timeseries.tsv"),
"space": self.space,
"preproc": "fmriprep",
}

Specifically, *space-MNI152NLin2009cAsym_desc-brain_mask.nii.gz (which follows BIDS convention) become overwritten with **space-MNI152NLin2009cAsym_desc-preproc_mask.nii.gz, a type of file that does not exist in the dset repository (and which does not follow BIDS conventions). I have to modify the mask_filepath column manually in the study' events dataframe before passing it to the segmenter to avoid missing file errors.

Steps to reproduce

N/A

Environment

• OS: ubunto
• Python version: 3.12.3
• neuralset version (pip show neuralset): 0.2.3
• Relevant package versions: nilearn 0.14.0