Table 2 Each of the image quality issues/artefacts is defined in terms of the image series and location defined, and the calculation of quantitative metric

A

Signal to noise ratio (SNR)

Pelvic

DWI – b900

Bilateral ROIs were defined over the gluteal muscle.

\(\frac{std(gluteal\ signal)}{mean(gluteal\ signal)}\)  

B

Anterior thoracic signal loss

Thoracic

DWI – b900

Bilateral ROIs were defined over the pectoral and paravertebral muscle.

\(\frac{mean(paravertebral\ signal)}{mean(pectoral\ signal)}\)  

C

Metal susceptibility artefacts

Anywhere

DWI – b50

The radiologist identified the location. The number of affected slices was observed manually and a measurement tool was used to measure the maximum extent in the A/P direction.

C1: No. affected slices

C2: Maximum extent in A/P direction (mm)

D

Fat suppression

Pelvic

DWI – b50

Bilateral ROIs are defined over the gluteal muscle and over the adjacent fat.

\(\frac{mean(fat\ signal)}{mean(gluteal\ signal)}\)  

E

Ghosting

E1: Brain

DWI – b50

A contour was defined around the surface of the brain and four ROIs were defined in the background (anterior, posterior, left and right).

100*\(\frac{\begin{array}{c}\left(top\ bg+bottom\ bg\right)-\\ (left\ bg+right\ bg)\end{array}}{2\ mean(brain\ signal)}\)  

E2: Pelvic

DWI – b50

Bilateral ROIs were defined over the gluteal muscle and three ROIs were defined in the background (anterior and in the top corners).

100*\(\frac{2\left(top\ bg\right)-(left\ bg+right\ bg)}{2\ mean(gluteal\ signal)}\)  

F

Geometric distortion

F1: Brain

DWI – b50

Dixon (water)

A contour was defined around the surface of the brain on both series

Hausdorff distance between the two contours

F2: Pelvic

DWI – b50

Dixon (water)

A contour was defined around the surface of the muscle on both series

Hausdorff distance between the two contours

G

Eddy current distortion

Pelvic

DWI – b50

DWI – b900

A contour was defined around the surface of the muscle on both series. The anterior half of the image was discarded to exclude the effect of respiratory motion and the laterally interior 30 cm region was used to exclude the difficult-to-define lateral regions.

Hausdorff distance between the two contours

H

Fat/water swaps

Anywhere

Dixon (water)

The radiologist identified the location.

No quantitative metric