QC procedures for n-back task fMRI?

Hi everyone,

I’m currently working with fMRI data from an n-back task and I’ve been thinking about quality control procedures specific to this type of paradigm.

As far as I can tell, there isn’t a well-established or standardized QC pipeline in the literature specifically tailored for n-back fMRI tasks, beyond general fMRI preprocessing and QC steps. Because of this, I would really appreciate hearing about your experiences and approaches.

For context, I have preprocessed my data using FSL, including standard steps (motion correction, spatial normalization, smoothing, etc.).

I have come across a few relevant papers:

  • Editorial: Demonstrating quality control (QC) procedures in fMRI

  • Efficient evaluation of the Open QC task fMRI dataset

  • Quality control in functional MRI studies with MRIQC and fMRIPrep

However, they generally suggest that there is no task-specific QC standard for n-back paradigms. Only one of them mentions concrete exclusion criteria, such as discarding participants with performance below 40% and framewise displacement (FD) greater than 0.9.

Given this, I’m unsure whether additional task-specific QC criteria should be applied, particularly considering aspects like:

  • Performance-related exclusions (e.g., accuracy thresholds)

  • Motion thresholds

  • Motion-related spikes

  • Signal to noise ratios

  • Inspection of activation patterns

I would be very interested to know:

  • What QC steps do you typically apply for n-back fMRI data?

  • Do you use any specific thresholds or criteria for excluding participants?

  • Any recommended papers, guidelines, or best practices?

Thanks in advance for your insights!