FAQ, Tips, and Tricks¶
Should I run quality control of my images before running fMRIPrep?¶
Yes. You should do so before any processing/analysis takes place.
Oftentimes (more often than we would like), images have fatal artifacts and problems.
Some exclusion criteria for data quality should be pre-specified before QC and any screening of the original data. Those exclusion criteria must be designed in agreement with the goals and challenges of the experimental design. For instance, when it is planned to run some cortical thickness analysis, images should be excluded even when they present the most subtle ghosts or other artifacts that may introduce biases in surface reconstruction. However, if the same artifactual data was planned to be used just as a reference for spatial normalization, some of those artifacts should be noted, but may not grant exclusion of the data.
When using publicly available datasets, an additional concern is that images may have gone through some kind of preprocessing (see next question).
What if I find some images have undergone some pre-processing already (e.g., my T1w image is already skull-stripped)?¶
These images imply an unknown level of preprocessing (e.g. was it already bias-field corrected?), which makes it difficult to decide on best-practices for further processing. Hence, supporting such images was considered very low priority for fMRIPrep. For example, see #707 and an illustration of downstream consequences in #939.
So for OpenFMRI, we’ve been excluding these subjects, and for user-supplied data, we would recommend reverting to the original, defaced, T1w images to ensure more uniform preprocessing.
My fMRIPrep run is hanging…¶
When running on Linux platforms (or containerized environments, because they are built around
Ubuntu), there is a Python bug that affects fMRIPrep that drives the Linux kernel to kill
processes as a response to running out of memory.
Depending on the process killed by the kernel, fMRIPrep may crash with a
error or hang indefinitely, depending on settings.
While we are working on finding a solution that does not run up against this bug, this may take some
This can be most easily resolved by allocating more memory to the process, if possible.
Please find more information regarding this error from discussions on NeuroStars:
Additionally, consider using the
--low-mem flag, which will make some memory optimizations at the cost of disk space in the working directory.