Comparing multiband and singleband EPI in NODDI at 3\xa0T: what are the implications for reproducibility and study sample sizes?

Abstract

ObjectiveThe reproducibility of Neurite orientation dispersion and density imaging (NODDI) metrics from time-saving multiband (MB) EPI compared with singleband (SB) has not been considered. This study aims to evaluate the reproducibility of NODDI parameters from SB and MB acquisitions, determine the agreement between acquisitions and estimate the sample sizes required to detect between-group change.MethodsBrain diffusion MRI data were acquired using SB and MB (acceleration factors 2 (MB2) and 3 (MB3)) on 8 healthy subjects on 2 separate visits. NODDI maps of isotropic volume fraction (FISO), neurite density (NDI) and orientation dispersion index (ODI) were estimated. Region-of-interest analysis was performed; variability across subjects and visits was measured using coefficients of variation (CoV). Intraclass correlation coefficient and Bland–Altman analysis were performed to assess reproducibility and detect any systematic bias between SB, MB2 and MB3. Power calculations were used to determine sample sizes required to detect group differences.ResultsBoth NDI and ODI were reproducible between visits; however, FISO was variable. All parameters were not reproducible across methods; a systematic bias was observed with the derived values decreasing as the MB factor increases. The number of subjects needed to detect a between-group change is not significantly different between methods; however, ODI needs considerably higher sample sizes than NDI.ConclusionsBoth SB and MB yield highly reproducible NDI and ODI measures, but direct comparison of these parameters between methods is complicated by systematic differences that exist between the two approaches.