Linearity vs. Distortion
Geometric distortion refers to the inaccurate representation of the true size of an object, or the displacement of displayed points within an image relative to their true location in a slice. These effects are named linearity and distortion.
Distortion = [100 x SD(distances)] / average (distances)
Linearity manifests as an inappropriate change in height or width, or both. Distortion may be seen where straight lines in an object appear curved, or where there is apparent rotation of the image.
Linearity = average (distances)
Distances refers to multiple measurements of a distance on an image, where each distance measured is distance (mm) = distance (pixels) x pixel dimension
and
Pixel Dimension = FOV (mm)/ matrix (pixels)
Further reading on this topic:
Books: MRI From Picture to Proton p209, ACR Quality Control Manual p80-81, IPEM Report 80 p24-29