Table 4 Summary table of quality assessment values.

From: Crowdsourced MRI quality metrics and expert quality annotations for training of humans and machines

Expert rating
Exclude Assigned to images that show quality defects that preclude any type of processing
Poor Assigned to images that, although presenting some quality problem, may tolerate some types of processing. For instance, a T1w image that may be used as the co-registration reference, but will probably generate biased cortical thickness measurements.
Acceptable Assigned to images that do not show any substantial issue that may preclude processing
Excellent Assigned to images without quality issues
A vector of boolean values corresponding to the following list of possible artifacts found in the image: • Head motion artifacts • Eye spillover through phase-encoding axis • Non-eye spillover through phase-encoding axis • Coil failure • Global noise • Local noise • Electromagnetic interference/perturbation • Problematic field-of-view prescription/Wrap-around • Aliasing ghosts • Other ghosts • Intensity non-uniformity • Temporal field variation • Reconstruction and postprocessing (e.g. denoising, defacing, resamplings) • Uncategorized artifact
  1. Annotations received through the feedback widget are stored in a separate database collecting one rating value and an array of artifacts present in the image.