Assessment of a diffusion phantom for quality assurance in brain microstructure diffusion MRI studies

Abstract

Diffusion-weighted imaging (DWI) is a key contrast mechanism in MRI which allows for the assessment of microstructural properties of brain tissues by measuring the displacement of water molecules. Several diffusion models, including the tensor (DTI), kurtosis (DKI), and neurite orientation dispersion and density imaging (NODDI), are commonly used in both research and clinical practice. However, there is currently no standardized method for validating the stability and repeatability of these models over time. This study evaluates the use of a DTI phantom as a standard reference for diffusion MRI model validation. The phantom, along with four healthy volunteers, was scanned repeatedly on different days to assess repeatability and stability. The acquired data were fitted to the diffusion models, with repeatability assessed in the phantom using the coefficient of variation (CoV), while stability in vivo was assessed using the repeatability coefficient (RC). The phantom was consecutively scanned eight times to investigate the impact of gradient coil heating on measurement consistency. Results showed that the phantom provided a highly reproducible reference, with CoVs below 5% across repeated and consecutive acquisitions, confirming the robustness of the diffusion models. In vivo, the low RCs indicated that the models remained stable over time, despite potential physiological variability. This study highlights the essential role of phantoms in diffusion MRI research, providing a reference framework for model validation. Future research will expand on this work to a multi-center study to assess inter-scanner variability, potentially incorporating the phantom into calibration protocols to standardize diffusion MRI measurements across different MRI systems.

Keywords: DWI/DTI/DKI; Diffusion modelling; NODDI; Phantom; Validation.

Fig. 1

DTI maps obtained in the fiber ring of the phantom (a) and in the brain (b). The maps show the FA on the left and the MD on the right.

Fig. 2

DKI and MSDKI maps obtained in the fiber ring of the phantom (a) and in the brain (b). From left to right the maps show the DKI AK, DKI MK, MSDKI MK and the MSDKI MD. Except for the MSDKI MD, all the indices are dimensionless and range from 0 to 1.

Fig. 3

NODDI maps obtained in the fiber ring of the phantom (a) and in the brain (b). From left to right the maps show the β-fraction, the ODI, the tissue volume fraction (v.f.), the intra-neurite v.f. and the MSE. All the indices are dimensionless and range from 0 to 1.

Fig. 4

ROIs used to extract the results from the quantitative maps. (A) 3D binary mask manually designed to extract the results in the phantom study. ROIs for the in vivo study are the Corpus Callosum (B), the Anterior and Posterior limbs of the Internal Capsule (C), the Thalamus (D), the Putamen (E) and the Caudate (F). The brain ROIs were defined in the MNI152 space.

Fig. 5

The phantom is composed of a fiber ring with uniform anisotropy at each position, which is embedded in a homogeneous medium of water and sodium chloride to mimic restricted anisotropic diffusion in brain tissue.