The new QA analysis tool provides instant results from MR QA measurements from a uniform phantom, suitable for any volume coil (e.g. head, knee).
Step 1: Scan a uniform phantom using the desired pulse sequence. We recommend using a simple spin echo, 256×256 matrix, TE 10-20 ms, TR 500-1000 ms, 5 mm slice thickness, single acquisition, FOV 250 mm for head coil. Avoid phase oversampling, partial Fourier, rectangular field-of-view, parallel imaging, AI reconstruction, filtering. Use as many slices as you wish, but have at least a 1mm gap. Note the FE and PE directions.
HINT: If using a water based phantom, allow sufficient time for the fluid to settle. We recommend positioning the phantom at the close of the day’s list for scanning first thing next day.
Record the magnet room temperature.
Step 2: On the desired image, adjust window and level to visualise the noise.
Step 3: Draw a circular region of interest (ROI) that includes 75% of the phantom cross section area. Enter the following into the calculator:
- ROI mean
- ROI standard deviation
- ROI maximum value
- ROI minimum value
- Number of pixels
Step 4: Draw up to 4 ROIs in the noise. Avoid areas of obvious artefact (e.g. ghosting), roll-off). Aim for at least 20×20 pixels per ROI. Enter the following into the calculator:
- ROI mean
- ROI standard deviation
- Number of pixels
Leave blank any unused cells- e.g. if you used 2 or 3 noise ROIs.
The calculator will show:
- Raw SNR
- SNR corrected using the mean noise
- SNR corrected using the standard deviation (SD) of the noise
- An error value – using larger noise ROIs will reduce this.
- The integral uniformity– this should be greater than or equal to 80% for a head coil.
- Ghost-noise ratio
- Ghost-signal ratio
- Ghost-signal ratio corrected for noise – this should be less than 2.5% for a spin echo sequence with TE 10-20 ms.
Step 5: Plot your chosen SNR value (SD-corrected SNR recommended) on a chart against date. Set limits of the mean plus or minus the error value. Any point (e.g. day 10) lying outside this range requires further investigation.
Further information: MRI from Picture to Proton (3rd edn.) chapter 11.