TEST-RETEST RELIABILITY OF FRACTIONAL ANISOTROPY IN 5-YEAR-OLDS

Abstract

Diffusion tensor imaging (DTI) has provided great insights to the microstructural features of developing brain and has been shown to be reliable in infants. However, the repeatability of the DTI scalars for older pediatric age groups has not been thoroughly addressed. In this study, DTI scans of 5-year-olds were used to investigate the test-retest reliability of three different measurements with both voxel-wise and region of interest (ROI) analysis. Out of 96 diffusion encoding directions, divided into three parts, 20 unique diffusion encoding directions were chosen per measurement from 48 subjects. Tract based spatial analysis (TBSS) was used to extract fractional anisotropy (FA) values from those images and using the FA values the repeatability of the measurements was assessed by intraclass correlation coefficient (ICC) and standard error of measurement (SEM). Overall, FA values had high repeatability both in voxel-based analysis (ICC>0.73) and ROI analysis (for non-skeletonized ROI type 88% of the ROI labels: ICC>0.75, for skeletonized ROI type 87% of the ROI labels: ICC>0.75). Using a skeleton in the ROI analysis did not contribute to the repeatability and the volume size was found to be a contributing factor for repeatability. Interscanner reliability as well as reliability measured by using different atlases are yet to be investigated in 5-year-old data.