Image Quality, Artifacts, and Quality Assurance
Quality metrics
Understand SNR, CNR, spatial resolution (voxel size, FOV/matrix), and how they trade off. Smaller voxels raise resolution but lower SNR; averaging raises SNR at time cost. Know image uniformity, geometric distortion, and slice thickness/position accuracy as QA parameters.
Common artifacts
Motion (ghosting, blurring), chemical shift (fat-water misregistration at frequency-encode direction), susceptibility (signal dropout near air/tissue or metal), wrap-around/aliasing (FOV smaller than anatomy), truncation, zipper, and RF-shading at high field. Understand causes and mitigations: swap phase/frequency direction, increase bandwidth, use fat suppression, saturate bands, increase FOV, or use spin echo to reduce susceptibility.
QA program
Daily, weekly, monthly, and annual tests using a phantom: SNR, uniformity, geometric accuracy, slice thickness, ghosting, and, for on-board systems, laser and couch alignment. Track trends against ACR/tolerance criteria. For RT, geometric fidelity matters more than aesthetic quality.
Traps
Questions may attribute chemical shift to the wrong direction - fat-water shift occurs along the frequency-encoding direction. Susceptibility is worst with gradient echo and near metal; spin echo refocuses static dephasing. Do not confuse wrap-around (FOV too small) with truncation (Gibbs ringing).
Rule of thumb
When distortion threatens target delineation, increase readout bandwidth and prefer spin-echo or distortion-corrected sequences; verify with a geometric phantom.
Worked example
If a 3 T scan shows bright/dark bands near a metal hip, suspect susceptibility; increasing bandwidth and using a spin-echo sequence reduces the artifact.