The Impact of Multiband and In-Plane Acceleration on White Matter Microstructure Analysis

Abstract

Accelerated imaging in diffusion MRI has been widely used to reduce scan time. This can be particularly important in reducing the burden in patients, such as those with mild cognitive impairment (MCI). However, the impact on reliability is not fully understood. Moreover, the impact on effect sizes in group comparisons has not been examined. We conducted a test–retest study of the impact of simultaneous multislice (SMS, also called multiband) and in-plane acceleration (IPA, also called phase acceleration) on reliability and effect sizes in diffusion imaging in MCI, healthy older adults, and young adults. We evaluated diffusion tensor imaging measures (fractional anisotropy, mean diffusivity, axial diffusivity, and radial diffusivity) and neurite orientation and dispersion measures (orientation dispersion, isotropic volume fraction, intracellular volume fraction) under no acceleration (S1P1), SMS = 3 with no in-plane acceleration (S3P1), SMS = 3 with IPA = 2 (S3P2), S6P1, and S6P2, with scan times varying from over 20 min in S1P1 to under 4 min in S6P2. In white matter voxels, the ranking of the accelerations with respect to intraclass correlations (ICCs) was S1P1 $\approx$ S3P1 $\ge$ S3P2 $>$ S6P1 $>$ S6P2, with ICCs in the good range across most DWI measures in S1P1, S3P1, and S3P2, moderate to good in S6P1, and poor to moderate in S6P2. In-plane acceleration did not improve ICC in areas of high susceptibility distortion. Acceleration significantly impacted the values of white matter microstructure with an overall trend of increase in fractional anisotropy and decrease in orientation dispersion with increasing multiband acceleration. In group comparisons, effect sizes tended to be similar across S1P1, S3P1, S3P2, and S6P1, including medium effect sizes in MCI versus healthy older adults and large effect sizes in young versus healthy older adults. Our results provide guidance regarding the costs of acceleration (reduced ICC from high acceleration) while also characterizing the benefits (S3P1 has similar reliability as S1P1 while requiring one third of the acquisition time, ROI-level group comparisons similar between S1P1, S3P1, S3P2, and S6P1). The overall high reliability and medium effect sizes of white matter microstructure measures with a moderate SMS factor indicates accelerated DWI can be used in developing biomarkers of neurological decline.