Use of the NESMA Filter to Improve Myelin Water Fraction Mapping with Brain MRI

REUSED: A deep neural network method for rapid whole-brain high-resolution myelin water fraction mapping from extremely under-sampled MRI

Changes in myelination are a cardinal feature of brain development and the pathophysiology of several central nervous system diseases, including multiple sclerosis and dementias. Advanced magnetic resonance imaging (MRI) methods have been developed to probe myelin content through the measurement of myelin water fraction (MWF). However, the prolonged data acquisition and post-processing times of current MWF mapping methods pose substantial hurdles to their clinical implementation. Recently, fast steady-state MRI sequences have been implemented to produce high-spatial resolution whole-brain MWF mapping within ∼20 min. Despite the subsequent significant advances in the inversion algorithm to derive MWF maps from steady-state MRI, the high-dimensional nature of such inversion does not permit further reduction of the acquisition time by data under-sampling. In this work, we present an unprecedented reduction in the computation (∼30 s) and the acquisition time (∼7 min) required for whole-brain high-resolution MWF mapping through a new Neural Network (NN)-based approach, named NN-Relaxometry of Extremely Under-SamplEd Data (NN-REUSED). Our analyses demonstrate virtually similar accuracy and precision in derived MWF values using NN-REUSED compared to results derived from the fully sampled reference method. The reduction in the acquisition and computation times represents a breakthrough toward clinically practical MWF mapping.

C-NODDI: a constrained NODDI model for axonal density and orientation determinations in cerebral white matter

Purpose

Neurite orientation dispersion and density imaging (NODDI) provides measures of neurite density and dispersion through computation of the neurite density index (NDI) and the orientation dispersion index (ODI). However, NODDI overestimates the cerebrospinal fluid water fraction in white matter (WM) and provides physiologically unrealistic high NDI values. Furthermore, derived NDI values are echo-time (TE)-dependent. In this work, we propose a modification of NODDI, named constrained NODDI (C-NODDI), for NDI and ODI mapping in WM.

Methods

Using NODDI and C-NODDI, we investigated age-related alterations in WM in a cohort of 58 cognitively unimpaired adults. Further, NDI values derived using NODDI or C-NODDI were correlated with the neurofilament light chain (NfL) concentration levels, a plasma biomarker of axonal degeneration. Finally, we investigated the TE dependence of NODDI or C-NODDI derived NDI and ODI.

Results

ODI derived values using both approaches were virtually identical, exhibiting constant trends with age. Further, our results indicated a quadratic relationship between NDI and age suggesting that axonal maturation continues until middle age followed by a decrease. This quadratic association was notably significant in several WM regions using C-NODDI, while limited to a few regions using NODDI. Further, C-NODDI-NDI values exhibited a stronger correlation with NfL concentration levels as compared to NODDI-NDI, with lower NDI values corresponding to higher levels of NfL. Finally, we confirmed the previous finding that NDI estimation using NODDI was dependent on TE, while NDI derived values using C-NODDI exhibited lower sensitivity to TE in WM.

Conclusion

C-NODDI provides a complementary method to NODDI for determination of NDI in white matter.

Span of regularization for solution of inverse problems with application to magnetic resonance relaxometry of the brain

We present a new regularization method for the solution of the Fredholm integral equation (FIE) of the first kind, in which we incorporate solutions corresponding to a range of Tikhonov regularizers into the end result. This method identifies solutions within a much larger function space, spanned by this set of regularized solutions, than is available to conventional regularization methods. An additional key development is the use of dictionary functions derived from noise-corrupted inversion of the discretized FIE. In effect, we combine the stability of solutions with greater degrees of regularization with the resolution of those that are less regularized. The span of regularizations (SpanReg) method may be widely applicable throughout the field of inverse problems.

Insights into human cerebral white matter maturation and degeneration across the adult lifespan

White matter (WM) microstructural properties change across the adult lifespan and with neuronal diseases. Understanding microstructural changes due to aging is paramount to distinguish them from neuropathological changes. Conducted on a large cohort of 147 cognitively unimpaired subjects, spanning a wide age range of 21 to 94 years, our study evaluated sex- and age-related differences in WM microstructure. Specifically, we used diffusion tensor imaging (DTI) magnetic resonance imaging (MRI) indices, sensitive measures of myelin and axonal density in WM, and myelin water fraction (MWF), a measure of the fraction of the signal of water trapped within the myelin sheets, to probe these differences. Furthermore, we examined regional correlations between MWF and DTI indices to evaluate whether the DTI metrics provide information complementary to MWF. While sexual dimorphism was, overall, nonsignificant, we observed region-dependent differences in MWF, that is, myelin content, and axonal density with age and found that both exhibit nonlinear, but distinct, associations with age. Furthermore, DTI indices were moderately correlated with MWF, indicating their good sensitivity to myelin content as well as to other constituents of WM tissue such as axonal density. The microstructural differences captured by our MRI metrics, along with their weak to moderate associations with MWF, strongly indicate the potential value of combining these outcome measures in a multiparametric approach. Furthermore, our results support the last-in-first-out and the gain-predicts-loss hypotheses of WM maturation and degeneration. Indeed, our results indicate that the posterior WM regions are spared from neurodegeneration as compared to anterior regions, while WM myelination follows a temporally symmetric time course across the adult life span.

Revisiting the T2 spectrum imaging inverse problem: Bayesian regularized non-negative least squares

Multi-echo T₂ magnetic resonance images contain information about the distribution of T₂ relaxation times of compartmentalized water, from which we can estimate relevant brain tissue properties such as the myelin water fraction (MWF). Regularized non-negative least squares (NNLS) is the tool of choice for estimating non-parametric T₂ spectra. However, the estimation is ill-conditioned, sensitive to noise, and highly affected by the employed regularization weight. The purpose of this study is threefold: first, we want to underline that the apparently innocuous use of two alternative parameterizations for solving the inverse problem, which we called the standard and alternative regularization forms, leads to different solutions; second, to assess the performance of both parameterizations; and third, to propose a new Bayesian regularized NNLS method (BayesReg). The performance of BayesReg was compared with that of two conventional approaches (L-curve and Chi-square (X²) fitting) using both regularization forms. We generated a large dataset of synthetic data, acquired in vivo human brain data in healthy participants for conducting a scan-rescan analysis, and correlated the myelin content derived from histology with the MWF estimated from ex vivo data. Results from synthetic data indicate that BayesReg provides accurate MWF estimates, comparable to those from L-curve and X², and with better overall stability across a wider signal-to-noise range. Notably, we obtained superior results by using the alternative regularization form. The correlations reported in this study are higher than those reported in previous studies employing the same ex vivo and histological data. In human brain data, the estimated maps from L-curve and BayesReg were more reproducible. However, the T₂ spectra produced by BayesReg were less affected by over-smoothing than those from L-curve. These findings suggest that BayesReg is a good alternative for estimating T₂ distributions and MWF maps.

Multidimensional MRI for Characterization of Subtle Axonal Injury Accelerated Using an Adaptive Nonlocal Multispectral Filter

Multidimensional MRI is an emerging approach that simultaneously encodes water relaxation (T₁ and T₂) and mobility (diffusion) and replaces voxel-averaged values with subvoxel distributions of those MR properties. While conventional (i.e., voxel-averaged) MRI methods cannot adequately quantify the microscopic heterogeneity of biological tissue, using subvoxel information allows to selectively map a specific T₁-T₂-diffusion spectral range that corresponds to a group of tissue elements. The major obstacle to the adoption of rich, multidimensional MRI protocols for diagnostic or monitoring purposes is the prolonged scan time. Our main goal in the present study is to evaluate the performance of a nonlocal estimation of multispectral magnitudes (NESMA) filter on reduced datasets to limit the total acquisition time required for reliable multidimensional MRI characterization of the brain. Here we focused and reprocessed results from a recent study that identified potential imaging biomarkers of axonal injury pathology from the joint analysis of multidimensional MRI, in particular voxelwise T₁-T₂ and diffusion-T₂ spectra in human Corpus Callosum, and histopathological data. We tested the performance of NESMA and its effect on the accuracy of the injury biomarker maps, relative to the co-registered histological reference. Noise reduction improved the accuracy of the resulting injury biomarker maps, while permitting data reduction of 35.7 and 59.6% from the full dataset for T₁-T₂ and diffusion-T₂ cases, respectively. As successful clinical proof-of-concept applications of multidimensional MRI are continuously being introduced, reliable and robust noise removal and consequent acquisition acceleration would advance the field towards clinically-feasible diagnostic multidimensional MRI protocols.

Decreased myelin content of the fornix predicts poorer memory performance beyond vascular risk, hippocampal volume, and fractional anisotropy in nondemented older adults

Alterations to cerebral white matter tracts have been associated with cognitive decline in aging and Alzheimer's disease (AD). In particular, the fornix has been implicated as especially vulnerable given that it represents the primary outflow tract of the hippocampus. Despite this, little work has focused on the fornix using a potential early marker of white matter degeneration-myelin water fraction (MWF; an in vivo marker of myelin content). Therefore, we sought to (1) clarify associations between MWF in the fornix and memory functioning, and (2) examine whether fornix MWF relates to memory performance above and beyond hippocampal volume and conventional imaging measures of white matter that may not be as specific to alterations in myelin content. Forty nondemented older adults (mean age = 72.9 years) underwent an MRI exam and neuropsychological assessment. Multicomponent driven equilibrium single pulse observation of T1 and T2 (mcDESPOT) was used to quantify fornix MWF and diffusion tensor imaging (DTI) was used to measure fornix fractional anisotropy (FA). Adjusting for age, sex, education, and vascular risk factors, linear regression models revealed that, lower fornix MWF was significantly associated with poorer memory functioning (β = 0.405, p = .007) across our sample of older adults. Notably, fornix MWF remained a significant predictor of memory functioning (β = 0.380, p = .015) even after adjusting for fornix DTI FA and hippocampal volume (in addition to the above covariates). Given the observed associations between myelin and memory in older adults without dementia, MWF may be a useful early marker of dementia risk.

Comparison of non-parametric T2 relaxometry methods for myelin water quantification

Multi-component T₂ relaxometry allows probing tissue microstructure by assessing compartment-specific T₂ relaxation times and water fractions, including the myelin water fraction. Non-negative least squares (NNLS) with zero-order Tikhonov regularization is the conventional method for estimating smooth T₂ distributions. Despite the improved estimation provided by this method compared to non-regularized NNLS, the solution is still sensitive to the underlying noise and the regularization weight. This is especially relevant for clinically achievable signal-to-noise ratios. In the literature of inverse problems, various well-established approaches to promote smooth solutions, including first-order and second-order Tikhonov regularization, and different criteria for estimating the regularization weight have been proposed, such as L-curve, Generalized Cross-Validation, and Chi-square residual fitting. However, quantitative comparisons between the available reconstruction methods for computing the T₂ distribution, and between different approaches for selecting the optimal regularization weight, are lacking. In this study, we implemented and evaluated ten reconstruction algorithms, resulting from the individual combinations of three penalty terms with three criteria to estimate the regularization weight, plus non-regularized NNLS. Their performance was evaluated both in simulated data and real brain MRI data acquired from healthy volunteers through a scan-rescan repeatability analysis. Our findings demonstrate the need for regularization. As a result of this work, we provide a list of recommendations for selecting the optimal reconstruction algorithms based on the acquired data. Moreover, the implemented methods were packaged in a freely distributed toolbox to promote reproducible research, and to facilitate further research and the use of this promising quantitative technique in clinical practice.

Model-informed machine learning for multi-component T2 relaxometry

Recovering the T₂ distribution from multi-echo T₂ magnetic resonance (MR) signals is challenging but has high potential as it provides biomarkers characterizing the tissue micro-structure, such as the myelin water fraction (MWF). In this work, we propose to combine machine learning and aspects of parametric (fitting from the MRI signal using biophysical models) and non-parametric (model-free fitting of the T₂ distribution from the signal) approaches to T₂ relaxometry in brain tissue by using a multi-layer perceptron (MLP) for the distribution reconstruction. For training our network, we construct an extensive synthetic dataset derived from biophysical models in order to constrain the outputs with a priori knowledge of in vivo distributions. The proposed approach, called Model-Informed Machine Learning (MIML), takes as input the MR signal and directly outputs the associated T₂ distribution. We evaluate MIML in comparison to a Gaussian Mixture Fitting (parametric) and Regularized Non-Negative Least Squares algorithms (non-parametric) on synthetic data, an ex vivo scan, and high-resolution scans of healthy subjects and a subject with Multiple Sclerosis. In synthetic data, MIML provides more accurate and noise-robust distributions. In real data, MWF maps derived from MIML exhibit the greatest conformity to anatomical scans, have the highest correlation to a histological map of myelin volume, and the best unambiguous lesion visualization and localization, with superior contrast between lesions and normal appearing tissue. In whole-brain analysis, MIML is 22 to 4980 times faster than the non-parametric and parametric methods, respectively.

Parsimonious discretization for characterizing multi-exponential decay in magnetic resonance

We address the problem of analyzing noise-corrupted magnetic resonance transverse decay signals as a superposition of underlying independently decaying monoexponentials of positive amplitude. First, we indicate the manner in which this is an ill-conditioned inverse problem, rendering the analysis unstable with respect to noise. Second, we define an approach to this analysis, stabilized solely by the nonnegativity constraint without regularization. This is made possible by appropriate discretization, which is coarser than that often used in practice. Thirdly, we indicate further stabilization by inspecting the plateaus of cumulative distributions. We demonstrate our approach through analysis of simulated myelin water fraction measurements, and compare the accuracy with more conventional approaches. Finally, we apply our method to brain imaging data obtained from a human subject, showing that our approach leads to maps of the myelin water fraction which are much more stable with respect to increasing noise than those obtained with conventional approaches.

Nonlinear associations of neurite density and myelin content with age revealed using multicomponent diffusion and relaxometry magnetic resonance imaging

Most magnetic resonance imaging (MRI) studies investigating the relationship between regional brain myelination or axonal density and aging have relied upon nonspecific methods to probe myelin and axonal content, including diffusion tensor imaging and relaxation time mapping. While these studies have provided pivotal insights into changes in cerebral architecture with aging and pathology, details of the underlying microstructural alterations have not been fully elucidated. In the current study, we used the BMC-mcDESPOT analysis, a direct and specific multicomponent relaxometry method for imaging of myelin water fraction (MWF), a marker of myelin content, and NODDI, an emerging multicomponent diffusion technique, for neurite density index (NDI) imaging, a proxy of axonal density. We investigated age-related differences in MWF and NDI in several white matter brain regions in a cohort of cognitively unimpaired participants over a wide age range. Our results indicate a quadratic, inverted U-shape, relationship between MWF and age in all brain regions investigated, suggesting that myelination continues until middle age followed by a decrease at older ages, in agreement with previous work. We found a similarly complex regional association between NDI and age, with several cerebral structures also exhibiting a quadratic, inverted U-shape, relationship. This novel observation suggests an increase in axonal density until the fourth decade of age followed by a rapid loss at older ages. We also observed that these age-related differences in MWF and NDI vary across different brain regions, as expected. Finally, our study indicates no significant association between MWF and NDI in most cerebral structures investigated, although this association approached significance in a limited number of brain regions, indicating the complementary nature of their information and encouraging further investigation. Overall, we find evidence of nonlinear associations between age and myelin or axonal density in a sample of well-characterized adults, using direct myelin and axonal content imaging methods.

Quantitative age-dependent differences in human brainstem myelination assessed using high-resolution magnetic resonance mapping

Previous in-vivo magnetic resonance imaging (MRI)-based studies of age-related differences in the human brainstem have focused on volumetric morphometry. These investigations have provided pivotal insights into regional brainstem atrophy but have not addressed microstructural age differences. However, growing evidence indicates the sensitivity of quantitative MRI to microstructural tissue changes in the brain. These studies have largely focused on the cerebrum, with very few MR investigations addressing age-dependent differences in the brainstem, in spite of its central role in the regulation of vital functions. Several studies indicate early brainstem alterations in a myriad of neurodegenerative diseases and dementias. The paucity of MR-focused investigations is likely due in part to the challenges imposed by the small structural scale of the brainstem itself as well as of substructures within, requiring accurate high spatial resolution imaging studies. In this work, we applied our recently developed approach to high-resolution myelin water fraction (MWF) mapping, a proxy for myelin content, to investigate myelin differences with normal aging within the brainstem. In this cross-sectional investigation, we studied a large cohort (n = 125) of cognitively unimpaired participants spanning a wide age range (21-94 years) and found a decrease in myelination with age in most brainstem regions studied, with several regions exhibiting a quadratic association between myelin and age. We believe that this study is the first investigation of MWF differences with normative aging in the adult brainstem. Further, our results provide reference MWF values.

Adult brain aging investigated using BMC-mcDESPOT-based myelin water fraction imaging

The relationship between regional brain myelination and aging has been the subject of intense study, with magnetic resonance imaging perhaps the most effective modality for elucidating this. However, most of these studies have used nonspecific methods to probe myelin content, including diffusion tensor imaging, magnetization transfer ratio, and relaxation times. In the present study, we used the BMC-mcDESPOT analysis, a direct and specific method for imaging of myelin water fraction (MWF), a surrogate of myelin content. We investigated age-related differences in MWF in several brain regions in a large cohort of cognitively unimpaired participants, spanning a wide age range. Our results indicate a quadratic, inverted U-shape, relationship between MWF and age in all brain regions investigated, suggesting that myelination continues until middle age followed by decreases at older ages. We also observed that these age-related differences vary across different brain regions, as expected. Our results provide evidence for nonlinear associations between age and myelin in a large sample of well-characterized adults, using a direct myelin content imaging method.

A simple and fast adaptive nonlocal multispectral filtering algorithm for efficient noise reduction in magnetic resonance imaging

PURPOSE

We recently introduced a multispectral (MS) nonlocal (NL) filter based on maximum likelihood estimation (MLE) of voxel intensities, termed MS-NLML. While MS-NLML provides excellent noise reduction and improved image feature preservation as compared to other NL or MS filters, it requires considerable processing time, limiting its application in routine analyses. In this work, we introduced a fast, simple, and robust filter, termed nonlocal estimation of multispectral magnitudes (NESMA), for noise reduction in multispectral (MS) magnetic resonance imaging (MRI).

METHODS

Through extensive simulation and in-vivo analyses, we compared the performance of NESMA and MS-NLML in terms of noise reduction and processing efficiency. Further, we introduce two simple adaptive methods that permit spatial variation of similar voxels, R, used in the filtering. The first method is semi-adaptive and permits variation of R across the image by using a relative Euclidean distance (RED) similarity threshold. The second method is fully adaptive and filters the raw data with several RED similarity thresholds to spatially determine the optimal threshold value using an unbiased criterion.

RESULTS

NESMA shows very similar filtering performance as compared to MS-NLML, however, with much simple implementation and very fast processing time. Further, for both filters, the adaptive methods were shown to further reduce noise in comparison with the conventional non-adaptive method in which R is set to a constant value throughout the image.

CONCLUSIONS

NESMA is fast, robust, and straightforward to implement filter. These features render it suitable for routine clinical use and analysis of large MRI datasets.

Spatially adaptive unsupervised multispectral nonlocal filtering for improved cerebral blood flow mapping using arterial spin labeling magnetic resonance imaging

BACKGROUND

Cerebral blood flow (CBF) is an emerging biomarker for normal aging and neurodegenerative diseases. Arterial spin labeling (ASL) perfusion MRI permits noninvasive quantification of CBF. However, high-quality mapping of CBF from ASL imaging is challenging, largely due to noise.

NEW METHOD

We demonstrate the ability of the recently introduced nonlocal estimation of multispectral magnitudes (NESMA) filter to greatly improve determination of CBF estimates from ASL imaging data. We evaluated the results of NESMA-ASL for CBF mapping from data obtained on human brain (n = 10) across a wide age range (21-74 years) using a standard clinical protocol. Results were compared to those obtained from unfiltered images or filtered images using conventional and advanced filters. Quantitative analyses for different spatial image resolutions and signal-to-noise ratios, SNRs, were also conducted.

RESULTS

Our results demonstrate the potential of NESMA-ASL to permit high-quality high-resolution CBF mapping. NESMA-ASL substantially reduces random variation in derived CBF estimates while preserving edges and small structures, with minimal bias and dispersion in derived CBF estimates.

COMPARISON WITH EXISTING METHODS

NESMA-ASL outperforms all evaluated filters in terms of noise reduction and detail preservation. Further, unlike other filters, NESMA-ASL is straightforward to implement requiring only one user-defined parameter, which is relatively insensitive to SNR or local image structure.

CONCLUSIONS

In-vivo estimation of CBF in the human brain from ASL imaging data was markedly improved through use of the NESMA-ASL filter. The use of NESMA-ASL may contribute significantly to the goal of high-quality high-resolution CBF mapping within a clinically feasible acquisition time.