Artifacts in Diffusion MRI

Abstract In addition to thermal noise, several artifacts can adversely affect the accuracy and precision of diffusion MRI measures. This chapter identifies the most relevant sources of artifacts in typical echo-planar diffusion MRI scans in a clinical setting. It examines the effects of image distortions originating from susceptibility differences, concomitant fields effects, and eddy currents on diffusion tensor computed quantities. It also reviews artifacts originating from “physiological noise” which includes subject motion, cardiac pulsation, and, possibly, respiratory motion. These artifacts not only affect the reproducibility of clinical diffusion tensor MRI (DTI) scans but most importantly could produce biased results. Physiological noise may not be equally distributed in different subjects and it may confound the assessment of differences between groups. For example, diffusion MRI differences found between healthy controls and patients may be due to artifacts originating from physiological noise, such as heart rate and subject motion, rather than to true anatomical differences. Fortunately, there are several data processing approaches that can successfully remove effects of artifacts and improve the quality of the computed diffusion metrics. The chapter reviews these approaches and we provide practical suggestions for designing acquisition protocols that help minimize the impact of artifacts originating from physiological noise and other sources.