Impact of MRI acceleration methods on Diffusion Tensor Imaging scalar metrics
4.58 MB
avoin
Julkaisu on tekijänoikeussäännösten alainen. Teosta voi lukea ja tulostaa henkilökohtaista käyttöä varten. Käyttö kaupallisiin tarkoituksiin on kielletty.
Lataukset32
Pysyvä osoite
Verkkojulkaisu
DOI
Tiivistelmä
Diffusion weighted imaging is an advanced magnetic resonance imaging technique (MRI) that is used to measure diffusion of free water molecules providing insights into tissue microstructure. This can provide complementary information to structural MRI regarding various diseases. Diffusion tensor imaging technique utilizes this method further by measuring multiple diffusion encoding directions and modelling this data to derive information on intra-voxel anisotropy. However, DTI acquisition is subject to long image acquisition times and confounding effects such as subject motion, hindering effective employment of technique in clinical imaging workflows. Inherent slow acquisition time of MRI images including DTI can be reduced by using various acceleration techniques such as parallel imaging (PI). Available acceleration techniques vary in function, falling generally into two categories: those that minimize acquisition time and those that optimize data through denoising and iterative sampling. Using acceleration methods is always subject to alter the MRI data with reason to assess reliability of acceleration technique before implementing them into wider clinical use. DTI is prone to confounding effects and long acquisition time required for meaningful data collection. In this study the impact of different acceleration techniques was evaluated in a multi-vendor and -scanner setting, where varying sampling schemes using various acceleration methods are compared to clinically used “state of the art” acquisition sample scheme intra-scanner. In this study, impact of various vendor-supplied MRI acceleration methods was assessed with diffusion tensor imaging and derived diffusion scalars in healthy volunteers using various MRI scanners from different vendors. Imaging data was pre-processed using same pipeline and diffusion scalars fractional anisotropy, axial diffusivity, mean diffusivity and radial diffusivity data were extracted. 48 Regions of Interests were extracted from subject specific white matter-skeleton derived using Tract Based Spatial Statistics. Mean ROI scalar-measurements were analyzed using intraclass correlation coefficient, Dice similarity coefficient, Coefficient of Variance and Bland-Altman plot. Measurements from accelerated sequences displayed good to excellent reliability in ICC, with comparable sampling schemes to reference sequences measuring negligible bias in Bland-Altman plot. Effect of DLIR on DTI scalars differed between two vendor-supplied algorithms.