Precision and accuracy in diffusion tensor magnetic resonance imaging

Abstract

This article reviews some of the key factors influencing the accuracy and precision of quantitative metrics derived from diffusion magnetic resonance imaging data. It focuses on the study pipeline beginning at the choice of imaging protocol, through preprocessing and model fitting up to the point of extracting quantitative estimates for subsequent analysis. The aim was to provide the newcomers to the field with sufficient knowledge of how their decisions at each stage along this process might impact on precision and accuracy, to design their study/approach, and to use diffusion tensor magnetic resonance imaging in the clinic. More specifically, emphasis is placed on improving accuracy and precision. I illustrate how careful choices along the way can substantially affect the sample size needed to make an inference from the data.