There are several things that can go wrong during the acquisition phase (scanning) of a subject, some of which severely impact the usability of a subject's dataset. While any given project (study) is still in the stage of data collection (subjects are still being scanned), there is always the chance to decide that a particular subject might introduce too much noise into the eventually performed group statistic and should be discarded (and in this case replaced by another subject).

Relevant for that decision could be one of the following issues:

• the subject had to exit the scanner before the experiment was completed → usually such a dataset needs to be discarded
• the subject couldn't restrain from moving their head during the experiment → depending on how difficult it is to find a replacement subject, it is advised to discard such a dataset
• the scanner produced disproportionally strong noise in the data → if possible, such a dataset should also be discarded

Of course there are still many other possible reasons to discard any given subject (e.g. a score on a questionnaire/behavioral measure indicates that the subject does not fall into the distribution of the examined population of subjects), but especially the second and third issue mentioned above can be detected even before entering a subject's dataset into any given group analysis.

To run the fMRI quality checking function, the images need to be in one of the functional imaging data formats currently supported by the xff class (Analyze/NIftI, FMR/STC, VTC).

The assessment is divided into two separate steps: one that performs several computational analysis and stores several results in a structure (which can be saved to disk for later), and a second step that assumes user interaction (i.e. manual inspection of the actual results of the computations).

Please consult the fmriquality reference manual page for all options and outputs.

The output of fmriquality can be passed on to fmriqasheet, which in turn creates a new figure and displays part of the information in the structure, which can be used to decide on whether or not a subject would likely introduce too much noise/bias at the group level.