mriqa

The mriqa module performs quality assurance checks on MR data. Every MR series has an entry in the mrqa table, and when the module runs, it checks the table to see which MR series do not have an entry, and if no entry exists, it runs the QA checks.

QA checks performed

  • Movement correction on 4D data: For any series that are 4D, movement correction is calculated using FSL. Volumes are aligned to the middle volume, and max/min displacement as well as total displacement are calculated. For example, if a subject moves 1mm in the positive X direction and 0.5mm in the negative X direction during the course of the series, the following will be recorded: max displacement: 1mm, min: 0.5mm, total: 1.5mm.
    Movement color coding: When the QA checks are displayed, the  values are color coded according to how “good” they are. Green indicates no movement, red indicates that the total displacement was greater than the voxel size in that dimension. For example, if the total displacement is 3.4mm in the X direction, but the  voxel size in the X direction was 3mm, the color will be red. Colors are displayed on a gradient from Green->Yellow->Orange->Red.
  • Signal to noise ratio: For all series, an inside-outside (IO) SNR is calculated. The 8 corners of the volume are considered to be the baseline and the center section considered the signal. The SNR is calculated from that. If the series is 4D, a second per-voxel (PV) SNR is calculated using the corners of the volume over time compared to the center region over time.
    Notes on the usefulness of SNR. SNR is NOT a useful QA measure to compare subjects to other subjects. One subject may have a low SNR value but a great quality image, and another subject may have a high SNR value, but a low quality image. SNR is only useful when comparing the same sequences within an imaging study for a particular subject. For example, if you run 5 T1 images, you may infer that higher a SNR for one series would imply it is the better image. You cannot infer that a higher SNR between subjects has any meaning. The chemistry of each subject alters the SNR value.
  • Thumbnail images: a thumbnail image of the middle slice of the volume is captured, colorized, an inverse FFT performed, and a radial average of the FFT is calculated. While not all slices are involved in this calculation, the resulting k-space like image may give insight into noise artifacts.

mrqa is set to run automatically via cron once per minute. To start the program manually type

cd /nidb/programs
perl mriqa.pl

Leave a Reply

Your email address will not be published. Required fields are marked *

*


one − 1 =

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>