Overview

The high-frequency MRI database is a repeated-measure study that collects multiple microstructural MRI data:

  • anatomical (3DT1 and FLAIR)

  • multi-shell diffusion weighted images (DWI)

  • inhomogeneous MT images (ihMT - MTI)

with a high number of MRI acquisitions over a short period of time in healthy subjects. This high-frequency dataset was designed to generate enough data per subject to optimize a relevant assessment of the consistency of different brain MRI measurements. The two aspects of consistency, reliability and variability, were evaluated using the Intraclass Coefficient correlation (ICC) value and within- and between-subject coefficient of variation (CV).

Acquisition scheme

Twenty healthy adults were scanned 5 times over 6 months at the Centre Hospitalier Universitaire of Sherbrooke (CHUS) using a clinical 3T MRI scanner (Ingenia, Philips Healthcare, Best, Netherlands).

Pipeline summary

The main steps of the analysis are summarized in the figure below.

../_images/pipeline_summary.png

MTI : Magnetization Transfer Imaging, DTI: Diffusion Tensor Imaging, NODDI: Neurite Orientation Density and Dispersion Imaging, HARDI : High Angular Resolution Imaging. For a list of metrics see Table of Metrics generated in Pipeline description.

Consistency summary

The main coherence results are summarized in the following figure :

../_images/summary_consistency.png

Classification of bundles and measures into four groups according to ICC values and within-variability (CVw). The x-axis represents CVw values (i.e., reproducibility), and the y-axis represents ICC values (i.e., reliability). High consistency group represents low CVw value < 0.05 (5%) and high ICC value > 0.75 (high reliability); Good consistency - ICC: high ICC value > 0.75 but high CVw value > 0.05 (>5%); Good consistency - CVw: low CVw value < 0.05 but low ICC value < 0.75 and finally, Moderate consistency: high CVw value > 0.05 (>5%) and low ICC value < 0.75.

Tools

The tools described and used are developed by the Sherbrooke Connectivity Imaging Lab (SCIL), see the SCIL Github page for all available nextflows and tools.