Comprehensive framework for accurate diffusion MRI parameter estimation

Abstract

During the last decade, many approaches have been proposed for improving the estimation of diffusion measures. These techniques have already shown an increase in accuracy based on theoretical considerations, such as incorporating prior knowledge of the data distribution. The increased accuracy of diffusion metric estimators is typically observed in well-defined simulations, where the assumptions regarding properties of the data distribution are known to be valid. In practice, however, correcting for subject motion and geometric eddy current deformations alters the data distribution tremendously such that it can no longer be expressed in a closed form. The image processing steps that precede the model fitting will render several assumptions on the data distribution invalid, potentially nullifying the benefit of applying more advanced diffusion estimators. In this work, we present a generic diffusion model fitting framework that considers some statistics of diffusion MRI data. A central role in the framework is played by the conditional least squares estimator. We demonstrate that the accuracy of that particular estimator can generally be preserved, regardless the applied preprocessing steps, if the noise parameter is known a priori. To fulfill that condition, we also propose an approach for the estimation of spatially varying noise levels.

One diffusion acquisition and different white matter models: How does microstructure change in human early development based on WMTI and NODDI?

“…The variation in spatial resolution between datasets caused some variability in SNR. However, the Rician bias was accounted for in the NODDI fitting procedure, and was found to be insignificant in the DKI fitting procedure (by comparing the current weighted linear least-squares algorithm with a conditional least-squares algorithm (Veraart et al, 2013a)). Moreover, because the difference in spatial resolution between the datasets was uncorrelated with the subject age, it is highly unlikely that these differences affected the trend of parameters with age highlighted in this study.…”

“…An in-house developed pipeline written in Matlab (MathWorks, Natick, MA, USA) was used for noise level estimation in each voxel (Veraart et al, 2013a), motion and eddy current correction (Ben-Amitay et al, 2012). The images were skull-stripped and a mask for CSF exclusion based on signal intensity in the b=0 image was created.…”

Choice of Diffusion Tensor Estimation Approach Affects Fiber Tractography of the Fornix in Preterm Brain

“…Thus, strategies to limit image corruption should be incorporated into setups to acquire neonatal DTI data 27 ; this includes 1) prevention of motion by comforting the infant and promoting natural sleep 28 ; 2) adjustment of parameter settings, by shortening diffusion time, applying stronger gradients, or by use of lower b-values 7 ; 3) oversampling gradient-sensitizing directions and removing corrupted diffusionweighted images 6 ; and 4) applying more advanced tensor estimation methods. 5 Because diffusion tensor estimation techniques differ considerably in principle, speed, and accuracy, 29 awareness of the benefits and pitfalls is essential: the LLS method is fast and mostly used but assumes that errors are identically distributed, which can result in inaccurate estimation of the tensor. 30 The WLLS method is slightly slower but provides more accurate results because it considers errors to be heterogeneously distributed.…”

Diffusion kurtosis imaging allows the early detection and longitudinal follow-up of amyloid-β-induced pathology

“…The diffusion tensor and kurtosis tensor quantify the Gaussian diffusion profile and the deviation from a Gaussian diffusion distribution, respectively [10,14]. The diffusion tensor and the diffusion kurtosis tensor were estimated simultaneously using conditional least squares estimators while imposing positivity on the kurtosis coefficients [15]. The conditional least squares estimator explicitly accounts for the Rician MR data distribution, for which the noise level has been estimated [16].…”