Adaptive phase correction of diffusion-weighted images

Brain fiber structure estimation based on principal component analysis and RINLM filter

Diffusion magnetic resonance imaging is a technique for non-invasive detection of microstructure in the white matter of the human brain, which is widely used in neuroscience research of the brain. However, diffusion-weighted images(DWI) are sensitive to noise, which affects the subsequent reconstruction of fiber orientation direction, microstructural parameter estimation and fiber tracking. In order to better eliminate the noise in diffusion-weighted images, this study proposes a noise reduction method combining Marchenko-Pastur principal component analysis(MPPCA) and rotation-invariant non-local means filter(RINLM) to further remove residual noise and preserve the image texture detail information. In this study, the algorithm is applied to the fiber structure and the prevailing microstructural models within the human brain voxels based on simulated and real human brain datasets. Experimental comparisons between the proposed method and the state-of-the-art methods are performed in single-fiber, multi-fiber, crossed and curved-fiber regions as well as in different microstructure estimation models. Results demonstrated the superior performance of the proposed method in denoising DWI data, which can reduce the angular error in fiber orientation reconstruction to obtain more valid fiber structure estimation and enable more complete fiber tracking trajectories with higher coverage. Meanwhile, the method reduces the estimation errors of various white matter microstructural parameters and verifies the performance of the method in white matter microstructure estimation.

Axial and radial axonal diffusivities and radii from single encoding strongly diffusion-weighted MRI

We enable the estimation of the per-axon axial diffusivity from single encoding, strongly diffusion-weighted, pulsed gradient spin echo data. Additionally, we improve the estimation of the per-axon radial diffusivity compared to estimates based on spherical averaging. The use of strong diffusion weightings in magnetic resonance imaging (MRI) allows to approximate the signal in white matter as the sum of the contributions from only axons. At the same time, spherical averaging leads to a major simplification of the modeling by removing the need to explicitly account for the unknown distribution of axonal orientations. However, the spherically averaged signal acquired at strong diffusion weightings is not sensitive to the axial diffusivity, which cannot therefore be estimated although needed for modeling axons - especially in the context of multi-compartmental modeling. We introduce a new general method for the estimation of both the axial and radial axonal diffusivities at strong diffusion weightings based on kernel zonal modeling. The method could lead to estimates that are free from partial volume bias with gray matter or other isotropic compartments. The method is tested on publicly available data from the MGH Adult Diffusion Human Connectome project. We report reference values of axonal diffusivities based on 34 subjects, and derive estimates of axonal radii from only two shells. The estimation problem is also addressed from the angle of the required data preprocessing, the presence of biases related to modeling assumptions, current limitations, and future possibilities.

Does perfect filtering really guarantee perfect phase correction for diffusion MRI data?

Owing to its merit of avoiding noise-floor, phase correction is recently used to reconstruct real-valued diffusion MRI data by employing an image filter to estimate the noise-free background phase. However, several studies report an unexpected signal-loss issue for their reconstruction results, with its causing reason still remaining unclear. Although phase correction has achieved promising results in mitigating the signal-loss issue via improving the employed image filter, we have observed counterintuitive results that an advanced filter generates severe artifacts in our previous work. Considering the potential issues with phase correction procedures, in this paper, we argue that even a perfect image filter is insufficient to produce perfect phase correction. To point out the reason why phase correction introduces signal-loss and address this issue, we first propose a complex polar coordinate system (CPCS) to analyze its procedures in detail; second, based on CPCS, we find that phase correction has not sufficiently utilized the background phase, and thus propose a quantitative criterion to fully exploit the background phase; eventually, we propose a phase calibration procedure to remedy current phase correction. Extensive experimental results, including those on synthetic and real diffusion MRI data, demonstrate that our proposed method significantly reduces signal-loss and also eliminates artifacts in FA maps, particularly with improved accuracy on FA.

Optimization of quasi-diffusion magnetic resonance imaging for quantitative accuracy and time-efficient acquisition

PURPOSE

Quasi-diffusion MRI (QDI) is a novel quantitative technique based on the continuous time random walk model of diffusion dynamics. QDI provides estimates of the diffusion coefficient, <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics><mml:mrow><mml:mspace/> <mml:msub><mml:mi>D</mml:mi> <mml:mrow><mml:mn>1</mml:mn> <mml:mo>,</mml:mo> <mml:mn>2</mml:mn></mml:mrow> </mml:msub> </mml:mrow> <mml:annotation>$$ {D}_{1,2} $$</mml:annotation></mml:semantics> </mml:math> in mm²  s^-1 and a fractional exponent, <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics><mml:mrow><mml:mi>α</mml:mi></mml:mrow> <mml:annotation>$$ \upalpha $$</mml:annotation></mml:semantics> </mml:math> , defining the non-Gaussianity of the diffusion signal decay. Here, the b-value selection for rapid clinical acquisition of QDI tensor imaging (QDTI) data is optimized.

METHODS

Clinically appropriate QDTI acquisitions were optimized in healthy volunteers with respect to a multi-b-value reference (MbR) dataset comprising 29 diffusion-sensitized images arrayed between <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics><mml:mrow><mml:mi>b</mml:mi> <mml:mo>=</mml:mo> <mml:mn>0</mml:mn></mml:mrow> <mml:annotation>$$ b=0 $$</mml:annotation></mml:semantics> </mml:math> and 5000 s mm^-2 . The effects of varying maximum b-value ( <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics> <mml:mrow><mml:msub><mml:mi>b</mml:mi> <mml:mi>max</mml:mi></mml:msub> </mml:mrow> <mml:annotation>$$ {b}_{\mathrm{max}} $$</mml:annotation></mml:semantics> </mml:math> ), number of b-value shells, and the effects of Rician noise were investigated.

RESULTS

QDTI measures showed <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics> <mml:mrow><mml:msub><mml:mi>b</mml:mi> <mml:mi>max</mml:mi></mml:msub> </mml:mrow> <mml:annotation>$$ {b}_{\mathrm{max}} $$</mml:annotation></mml:semantics> </mml:math> dependence, most significantly for <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics><mml:mrow><mml:mi>α</mml:mi></mml:mrow> <mml:annotation>$$ \upalpha $$</mml:annotation></mml:semantics> </mml:math> in white matter, which monotonically decreased with higher <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics> <mml:mrow><mml:msub><mml:mi>b</mml:mi> <mml:mi>max</mml:mi></mml:msub> </mml:mrow> <mml:annotation>$$ {b}_{\mathrm{max}} $$</mml:annotation></mml:semantics> </mml:math> leading to improved tissue contrast. Optimized 2 b-value shell acquisitions showed small systematic differences in QDTI measures relative to MbR values, with overestimation of <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics><mml:mrow><mml:mspace/> <mml:mspace/> <mml:msub><mml:mi>D</mml:mi> <mml:mrow><mml:mn>1</mml:mn> <mml:mo>,</mml:mo> <mml:mn>2</mml:mn></mml:mrow> </mml:msub> </mml:mrow> <mml:annotation>$$ \kern0.50em {D}_{1,2} $$</mml:annotation></mml:semantics> </mml:math> and underestimation of <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics><mml:mrow><mml:mi>α</mml:mi></mml:mrow> <mml:annotation>$$ \upalpha $$</mml:annotation></mml:semantics> </mml:math> in white matter, and overestimation of <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics> <mml:mrow><mml:msub><mml:mi>D</mml:mi> <mml:mrow><mml:mn>1</mml:mn> <mml:mo>,</mml:mo> <mml:mn>2</mml:mn></mml:mrow> </mml:msub> </mml:mrow> <mml:annotation>$$ {D}_{1,2} $$</mml:annotation></mml:semantics> </mml:math> and <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics><mml:mrow><mml:mi>α</mml:mi></mml:mrow> <mml:annotation>$$ \upalpha $$</mml:annotation></mml:semantics> </mml:math> anisotropies in gray and white matter. Additional shells improved the accuracy, precision, and reliability of QDTI estimates with 3 and 4 shells at <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics> <mml:mrow><mml:msub><mml:mi>b</mml:mi> <mml:mi>max</mml:mi></mml:msub> <mml:mo>=</mml:mo> <mml:mn>5000</mml:mn></mml:mrow> <mml:annotation>$$ {b}_{\mathrm{max}}=5000 $$</mml:annotation></mml:semantics> </mml:math>  s mm^-2 , and 4 b-value shells at <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics> <mml:mrow><mml:msub><mml:mi>b</mml:mi> <mml:mi>max</mml:mi></mml:msub> <mml:mo>=</mml:mo> <mml:mn>3960</mml:mn></mml:mrow> <mml:annotation>$$ {b}_{\mathrm{max}}=3960 $$</mml:annotation></mml:semantics> </mml:math>  s mm^-2 , providing minimal bias in <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics> <mml:mrow><mml:msub><mml:mi>D</mml:mi> <mml:mrow><mml:mn>1</mml:mn> <mml:mo>,</mml:mo> <mml:mn>2</mml:mn></mml:mrow> </mml:msub> </mml:mrow> <mml:annotation>$$ {D}_{1,2} $$</mml:annotation></mml:semantics> </mml:math> and <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:semantics><mml:mrow><mml:mi>α</mml:mi></mml:mrow> <mml:annotation>$$ \upalpha $$</mml:annotation></mml:semantics> </mml:math> compared to the MbR.

CONCLUSION

A highly detailed optimization of non-Gaussian dMRI for in vivo brain imaging was performed. QDI provided robust parameterization of non-Gaussian diffusion signal decay in clinically feasible imaging times with high reliability, accuracy, and precision of QDTI measures.

MR image reconstruction from undersampled data for image-guided radiation therapy using a patient-specific deep manifold image prior

Introduction

Recent advancements in radiotherapy (RT) have allowed for the integration of a Magnetic Resonance (MR) imaging scanner with a medical linear accelerator to use MR images for image guidance to position tumors against the treatment beam. Undersampling in MR acquisition is desired to accelerate the imaging process, but unavoidably deteriorates the reconstructed image quality. In RT, a high-quality MR image of a patient is available for treatment planning. In light of this unique clinical scenario, we proposed to exploit the patient-specific image prior to facilitate high-quality MR image reconstruction.

Methods

Utilizing the planning MR image, we established a deep auto-encoder to form a manifold of image patches of the patient. The trained manifold was then incorporated as a regularization to restore MR images of the same patient from undersampled data. We performed a simulation study using a patient case, a real patient study with three liver cancer patient cases, and a phantom experimental study using data acquired on an in-house small animal MR scanner. We compared the performance of the proposed method with those of the Fourier transform method, a tight-frame based Compressive Sensing method, and a deep learning method with a patient-generic manifold as the image prior.

Results

In the simulation study with 12.5% radial undersampling and 15% increase in noise, our method improved peak-signal-to-noise ratio by 4.46dB and structural similarity index measure by 28% compared to the patient-generic manifold method. In the experimental study, our method outperformed others by producing reconstructions of visually improved image quality.

What's new and what's next in diffusion MRI preprocessing

Diffusion MRI (dMRI) provides invaluable information for the study of tissue microstructure and brain connectivity, but suffers from a range of imaging artifacts that greatly challenge the analysis of results and their interpretability if not appropriately accounted for. This review will cover dMRI artifacts and preprocessing steps, some of which have not typically been considered in existing pipelines or reviews, or have only gained attention in recent years: brain/skull extraction, B-matrix incompatibilities w.r.t the imaging data, signal drift, Gibbs ringing, noise distribution bias, denoising, between- and within-volumes motion, eddy currents, outliers, susceptibility distortions, EPI Nyquist ghosts, gradient deviations, B1 bias fields, and spatial normalization. The focus will be on "what's new" since the notable advances prior to and brought by the Human Connectome Project (HCP), as presented in the predecessing issue on "Mapping the Connectome" in 2013. In addition to the development of novel strategies for dMRI preprocessing, exciting progress has been made in the availability of open source tools and reproducible pipelines, databases and simulation tools for the evaluation of preprocessing steps, and automated quality control frameworks, amongst others. Finally, this review will consider practical considerations and our view on "what's next" in dMRI preprocessing.

Does powder averaging remove dispersion bias in diffusion MRI diameter estimates within real 3D axonal architectures?

Noninvasive estimation of axon diameter with diffusion MRI holds the potential to investigate the dynamic properties of the brain network and pathology of neurodegenerative diseases. Recent studies use powder averaging to account for complex white matter architectures, but these have not been validated for real axonal geometries from regions that contain fibre crossings. Here, we present 120-304μm long segmented axons from X-ray nano-holotomography volumes of a splenium and crossing fibre region of a vervet monkey brain. We show that the axons in the complex crossing fibre region, which contains callosal, association, and corticospinal connections, are larger and exhibit a wider distribution than those of the splenium region. To accurately estimate the axon diameter in these regions, therefore, sensitivity to a wide range of diameters is required. We demonstrate how the q-value, b-value, signal-to-noise ratio and the assumed intra-axonal parallel diffusivity influence the range of measurable diameters with powder average approaches. Furthermore, we show how Gaussian distributed noise results in a wider range of measurable diameter at high b-values than Rician distributed noise, even at high signal-to-noise ratios of 100. The number of gradient directions is also shown to impose a lower bound on measurable diameter. Our results indicate that axon diameter estimation can be performed with only few b-shells, and that additional shells do not improve the accuracy of the estimate. For strong gradients available on human Connectom and preclinical scanners, Monte Carlo simulations of diffusion confirm that powder averaging techniques succeed in providing accurate estimates of axon diameter across a range of sequence parameters and diffusion times, even in complex white matter architectures. At relatively low b-values, the diameter estimate becomes sensitive to axonal microdispersion and the intra-axonal parallel diffusivity shows time dependency at both in vivo and ex vivo intrinsic diffusivities.

Noise Mapping and Removal in Complex-Valued Multi-Channel MRI via Optimal Shrinkage of Singular Values

In magnetic resonance imaging (MRI), noise is a limiting factor for higher spatial resolution and a major cause of prolonged scan time, owing to the need for repeated scans. Improving the signal-to-noise ratio is therefore key to faster and higher-resolution MRI. Here we propose a method for mapping and reducing noise in MRI by leveraging the inherent redundancy in complex-valued multi-channel MRI data. Our method leverages a provably optimal strategy for shrinking the singular values of a data matrix, allowing it to outperform state-of-the-art methods such as Marchenko-Pastur PCA in noise reduction. Our method reduces the noise floor in brain diffusion MRI by 5-fold and remarkably improves the contrast of spiral lung ¹⁹F MRI. Our framework is fast and does not require training and hyper-parameter tuning, therefore providing a convenient means for improving SNR in MRI.

Measuring compartmental T2-orientational dependence in human brain white matter using a tiltable RF coil and diffusion-T2 correlation MRI

The anisotropy of brain white matter microstructure manifests itself in orientational-dependence of various MRI contrasts, and can result in significant quantification biases if ignored. Understanding the origins of this orientation-dependence could enhance the interpretation of MRI signal changes in development, ageing and disease and ultimately improve clinical diagnosis. Using a novel experimental setup, this work studies the contributions of the intra- and extra-axonal water to the orientation-dependence of one of the most clinically-studied parameters, apparent transverse relaxation T2. Specifically, a tiltable receive coil is interfaced with an ultra-strong gradient MRI scanner to acquire multidimensional MRI data with an unprecedented range of acquisition parameters. Using this setup, compartmental T2 can be disentangled based on differences in diffusional-anisotropy, and its orientation-dependence further elucidated by re-orienting the head with respect to the main magnetic field B→0. A dependence of (compartmental) T2 on the fibre orientation w.r.t. B→0 was observed, and further quantified using characteristic representations for susceptibility- and magic angle effects. Across white matter, anisotropy effects were dominated by the extra-axonal water signal, while the intra-axonal water signal decay varied less with fibre-orientation. Moreover, the results suggest that the stronger extra-axonal T2 orientation-dependence is dominated by magnetic susceptibility effects (presumably from the myelin sheath) while the weaker intra-axonal T2 orientation-dependence may be driven by a combination of microstructural effects. Even though the current design of the tiltable coil only offers a modest range of angles, the results demonstrate an overall effect of tilt and serve as a proof-of-concept motivating further hardware development to facilitate experiments that explore orientational anisotropy. These observations have the potential to lead to white matter microstructural models with increased compartmental sensitivity to disease, and can have direct consequences for longitudinal and group-wise T2- and diffusion-MRI data analysis, where the effect of head-orientation in the scanner is commonly ignored.

Magic DIAMOND: Multi-fascicle diffusion compartment imaging with tensor distribution modeling and tensor-valued diffusion encoding

Diffusion tensor imaging provides increased sensitivity to microstructural tissue changes compared to conventional anatomical imaging but also presents limited specificity. To tackle this problem, the DIAMOND model subdivides the voxel content into diffusion compartments and draws from diffusion-weighted data to estimate compartmental non-central matrix-variate Gamma distributions of diffusion tensors. It models each sub-voxel fascicle separately, resolving crossing white-matter pathways and allowing for a fascicle-element (fixel) based analysis of microstructural features. Alternatively, specific features of the intra-voxel diffusion tensor distribution can be selectively measured using tensor-valued diffusion-weighted acquisition schemes. However, the impact of such schemes on estimating brain microstructural features has only been studied in a handful of parametric single-fascicle models. In this work, we derive a general Laplace transform for the non-central matrix-variate Gamma distribution, which enables the extension of DIAMOND to tensor-valued encoded data. We then evaluate this "Magic DIAMOND" model in silico and in vivo on various combinations of tensor-valued encoded data. Assessing uncertainty on parameter estimation via stratified bootstrap, we investigate both voxel-based and fixel-based metrics by carrying out multi-peak tractography. We demonstrate using in silico evaluations that tensor-valued diffusion encoding significantly improves Magic DIAMOND's accuracy. Most importantly, we show in vivo that our estimated metrics can be robustly mapped along tracks across regions of fiber crossing, which opens new perspectives for tractometry and microstructure mapping along specific white-matter tracts.

Axon diameter index estimation independent of fiber orientation distribution using high-gradient diffusion MRI

Axon diameter mapping using high-gradient diffusion MRI has generated great interest as a noninvasive tool for studying trends in axonal size in the human brain. One of the main barriers to mapping axon diameter across the whole brain is accounting for complex white matter fiber configurations (e.g., crossings and fanning), which are prevalent throughout the brain. Here, we present a framework for generalizing axon diameter index estimation to the whole brain independent of the underlying fiber orientation distribution using the spherical mean technique (SMT). This approach is shown to significantly benefit from the use of real-valued diffusion data with Gaussian noise, which reduces the systematic bias in the estimated parameters resulting from the elevation of the noise floor when using magnitude data with Rician noise. We demonstrate the feasibility of obtaining whole-brain orientationally invariant estimates of axon diameter index and relative volume fractions in six healthy human volunteers using real-valued diffusion data acquired on a dedicated high-gradient 3-Tesla human MRI scanner with 300 mT/m maximum gradient strength. The trends in axon diameter index are consistent with known variations in axon diameter from histology and demonstrate the potential of this generalized framework for revealing coherent patterns in axonal structure throughout the living human brain. The use of real-valued diffusion data provides a viable solution for eliminating the Rician noise floor and should be considered for all spherical mean approaches to microstructural parameter estimation.