Longitudinal assessment of recovery after spinal cord injury with behavioral measures and diffusion, quantitative magnetization transfer and functional magnetic resonance imaging.

Spinal cord injuries (SCIs) are a leading cause of disability and can severely impact the quality of life. However, to date, the processes of spontaneous repair of damaged spinal cord remain incompletely understood, partly due to a lack of appropriate longitudinal tracking methods. Noninvasive, multiparametric magnetic resonance imaging (MRI) provides potential biomarkers for the comprehensive evaluation of spontaneous repair after SCI. In this study in rats, a clinically relevant contusion injury was introduced at the lumbar level that impairs both hindlimb motor and sensory functions. Quantitative MRI measurements were acquired at baseline and serially post-SCI for up to 2 wk. The progressions of injury and spontaneous recovery in both white and gray matter were tracked longitudinally using pool-size ratio (PSR) measurements derived from quantitative magnetization transfer (qMT) methods, measurements of water diffusion parameters using diffusion tensor imaging (DTI) and intrasegment functional connectivity derived from resting state functional MRI. Changes in these quantitative imaging measurements were correlated with behavioral readouts. We found (a) a progressive decrease in PSR values within 2 wk post-SCI, indicating a progressive demyelination at the center of the injury that was validated with histological staining, (b) PSR correlated closely with fractional anisotropy and transverse relaxation of free water, but did not show significant correlations with behavioral recovery, and (c) preliminary evidence that SCI induced a decrease in functional connectivity between dorsal horns below the injury site at 24 h. Findings from this study not only confirm the value of qMT and DTI methods for assessing the myelination state of injured spinal cord but indicate that they may also have further implications on whether therapies targeted towards remyelination may be appropriate. Additionally, a better understanding of changes after SCI provides valuable information to guide and assess interventions.

Enabling Multi-Shell b-Value Generalizability of Data-Driven Diffusion Models with Deep SHORE.

Intra-voxel models of the diffusion signal are essential for interpreting organization of the tissue environment at micrometer level with data at millimeter resolution. Recent advances in data driven methods have enabled direct comparison and optimization of methods for in-vivo data with externally validated histological sections with both 2-D and 3-D histology. Yet, all existing methods make limiting assumptions of either (1) model-based linkages between b-values or (2) limited associations with single shell data. We generalize prior deep learning models that used single shell spherical harmonic transforms to integrate the recently developed simple harmonic oscillator reconstruction (SHORE) basis. To enable learning on the SHORE manifold, we present an alternative formulation of the fiber orientation distribution (FOD) object using the SHORE basis while representing the observed diffusion weighted data in the SHORE basis. To ensure consistency of hyper-parameter optimization for SHORE, we present our Deep SHORE approach to learn on a data-optimized manifold. Deep SHORE is evaluated with eight-fold cross-validation of a preclinical MRI-histology data with four b-values. Generalizability of in-vivo human data is evaluated on two separate 3T MRI scanners. Specificity in terms of angular correlation (ACC) with the preclinical data improved on single shell: 0.78 relative to 0.73 and 0.73, multi-shell: 0.80 relative to 0.74 (p < 0.001). In the in-vivo human data, Deep SHORE was more consistent across scanners with 0.63 relative to other multi-shell methods 0.39, 0.52 and 0.57 in terms of ACC. In conclusion, Deep SHORE is a promising method to enable data driven learning with DW-MRI under conditions with varying b-values, number of diffusion shells, and gradient directions per shell.

Harmonizing 1.5T/3T Diffusion Weighted MRI through Development of Deep Learning Stabilized Microarchitecture Estimators.

Diffusion weighted magnetic resonance imaging (DW-MRI) is interpreted as a quantitative method that is sensitive to tissue microarchitecture at a millimeter scale. However, the sensitization is dependent on acquisition sequences (e.g., diffusion time, gradient strength, etc.) and susceptible to imaging artifacts. Hence, comparison of quantitative DW-MRI biomarkers across field strengths (including different scanners, hardware performance, and sequence design considerations) is a challenging area of research. We propose a novel method to estimate microstructure using DW-MRI that is robust to scanner difference between 1.5T and 3T imaging. We propose to use a null space deep network (NSDN) architecture to model DW-MRI signal as fiber orientation distributions (FOD) to represent tissue microstructure. The NSDN approach is consistent with histologically observed microstructure (on previously acquired ex vivo squirrel monkey dataset) and scan-rescan data. The contribution of this work is that we incorporate identical dual networks (IDN) to minimize the influence of scanner effects via scan-rescan data. Briefly, our estimator is trained on two datasets. First, a histology dataset was acquired on three squirrel monkeys with corresponding DW-MRI and confocal histology (512 independent voxels). Second, 37 control subjects from the Baltimore Longitudinal Study of Aging (67-95 y/o) were identified who had been scanned at 1.5T and 3T scanners (b-value of 700 s/mm2, voxel resolution at 2.2mm, 30-32 gradient volumes) with an average interval of 4 years (standard deviation 1.3 years). After image registration, we used paired white matter (WM) voxels for 17 subjects and 440 histology voxels for training and 20 subjects and 72 histology voxels for testing. We compare the proposed estimator with super-resolved constrained spherical deconvolution (CSD) and a previously presented regression deep neural network (DNN). NSDN outperformed CSD and DNN in angular correlation coefficient (ACC) 0.81 versus 0.28 and 0.46, mean squared error (MSE) 0.001 versus 0.003 and 0.03, and general fractional anisotropy (GFA) 0.05 versus 0.05 and 0.09. Further validation and evaluation with contemporaneous imaging are necessary, but the NSDN is promising avenue for building understanding of microarchitecture in a consistent and device-independent manner.

Lipopolysaccharide Induced Opening of the Blood Brain Barrier on Aging 5XFAD Mouse Model.

The development of neurotherapeutics for many neurodegenerative diseases has largely been hindered by limited pharmacologic penetration across the blood-brain barrier (BBB). Previous attempts to target and clear amyloid-ß (Aß) plaques, a key mediator of neurodegenerative changes in Alzheimer's disease (AD), have had limited clinical success due to low bioavailability in the brain because of the BBB. Here we test the effects of inducing an inflammatory response to disrupt the BBB in the 5XFAD transgenic mouse model of AD. Lipopolysaccharide (LPS), a bacterial endotoxin recognized by the innate immune system, was injected at varying doses. 24 hours later, mice were injected with either thioflavin S, a fluorescent Aß-binding small molecule or 30 nm superparamagnetic iron oxide (SPIO) nanoparticles, both of which are unable to penetrate the BBB under normal physiologic conditions. Our results showed that when pretreated with 3.0 mg/kg LPS, thioflavin S can be found in the brain bound to Aß plaques in aged 5XFAD transgenic mice. Following the same LPS pretreatment, SPIO nanoparticles could also be found in the brain. However, when done on wild type or young 5XFAD mice, limited SPIO was detected. Our results suggest that the BBB in aged 5XFAD mouse model is susceptible to increased permeability mediated by LPS, allowing for improved delivery of the small molecule thioflavin S to target Aß plaques and SPIO nanoparticles, which are significantly larger than antibodies used in clinical trials for immunotherapy of AD. Although this approach demonstrated efficacy for improved delivery to the brain, LPS treatment resulted in significant weight loss even at low doses, resulting from the induced inflammatory response. These findings suggest inducing inflammation can improve delivery of small and large materials to the brain for improved therapeutic or diagnostic efficacy. However, this approach must be balanced with the risks of systemic inflammation.

The radial diffusivity and magnetization transfer pool size ratio are sensitive markers for demyelination in a rat model of type III multiple sclerosis (MS) lesions.

Determining biophysical sensitivity and specificity of quantitative magnetic resonance imaging is essential to develop effective imaging metrics of neurodegeneration. Among these metrics, apparent pool size ratio (PSR) from quantitative magnetization transfer (qMT) imaging and radial diffusivity (RD) from diffusion tensor imaging (DTI) are both known to relate to histological measure of myelin density and integrity. However their relative sensitivities towards quantitative myelin detection are unknown. In this study, we correlated high-resolution quantitative magnetic resonance imaging measures of subvoxel tissue structures with corresponding quantitative myelin histology in a lipopolysaccharide (LPS) mediated animal model of MS. Specifically, we acquired quantitative magnetization transfer (qMT) and diffusion tensor imaging (DTI) metrics (on the same tissue sample) in an animal model system of type III oligodendrogliopathy which lacked prominent lymphocytic infiltration, a system that had not been previously examined with quantitative MRI. We find that the qMT measured apparent pool size ratio (PSR) showed the strongest correlation with a histological measure of myelin content. DTI measured RD showed the next strongest correlation, and other DTI and relaxation parameters (such as the longitudinal relaxation rate (R1f) or fractional anisotropy (FA)) showed considerably weaker correlations with myelin content.

A new method for detecting exchanging amide protons using chemical exchange rotation transfer.

In this study, we introduce a new method for amide proton transfer imaging based on chemical exchange rotation transfer. It avoids several artifacts that plague conventional chemical exchange saturation transfer approaches by creating label and reference scans based on varying the irradiation pulse rotation angle (π and 2π radians) instead of the frequency offset (3.5 and -3.5 ppm). Specifically, conventional analysis is sensitive to confounding contributions from magnetic field (B(0)) inhomogeneities and, more problematically, inherently asymmetric macromolecular resonances. In addition, the lipid resonance at -3.5 ppm complicates the interpretation of the reference scan and decreases the resulting contrast. Finally, partial overlap of the amide signal by nearby amines and hydroxyls obscure the results. By avoiding these issues, our new method is a promising approach for imaging endogenous protein and peptide content and mapping pH.

Multi-angle ratiometric approach to measure chemical exchange in amide proton transfer imaging.

Amide proton transfer imaging, a specific form of chemical exchange saturation transfer imaging, has previously been applied to studies of acute ischemic acidosis, stroke, and cancer. However, interpreting the resulting contrast is complicated by its dependence on the exchange rate between amides and water, the amide concentration, amide and water relaxation, and macromolecular magnetization transfer. Hence, conventional chemical exchange saturation transfer contrast is not specific to changes such as reductions in pH due to tissue acidosis. In this article, a multi-angle ratiometric approach based on several pulsed-chemical exchange saturation transfer scans at different irradiation flip angles is proposed to specifically reflect exchange rates only. This separation of exchange effects in pulsed-chemical exchange saturation transfer experiments is based on isolating rotation vs. saturation contributions, and such methods form a new subclass of chemical exchange rotation transfer (CERT) experiments. Simulations and measurements of creatine/agar phantoms indicate that a newly proposed imaging metric isolates the effects of exchange rate changes, independent of other sample parameters.

Optimizing pulsed-chemical exchange saturation transfer imaging sequences.

Chemical exchange saturation transfer (CEST) provides a new imaging contrast mechanism sensitive to labile proton exchange. Pulsed-CEST imaging is better suited to the hardware constraints on clinical imaging systems when compared with traditional continuous wave-CEST imaging methods. However, designing optimum pulsed-CEST imaging sequences entails complicated and time-consuming numerical integrations. In this work, a simplified and computationally efficient technique is provided to optimize the pulsed-CEST imaging sequence. An analysis was performed of the optimal average irradiation power and the optimal irradiation flip angle as a function of the acquisition parameters and sample properties in both a two-pool model and a three-pool model of endogenous amine exchange. Key simulated and experimental results based on a creatine/agar tissue phantom show that (1) the average irradiation power is a more meaningful sequence metric than is the average irradiation field amplitude, (2) the optimal average powers for continuous wave and pulsed-CEST imaging are approximately equal to each other for a relevant range of solute frequency offsets, exchange rates, and concentrations, (3) an irradiation flip angle of 180° is optimal or near optimal, independent of the other acquisition parameters and the sample properties, and (4) higher duty cycles yield higher CEST contrast.

Optimized inversion recovery sequences for quantitative T1 and magnetization transfer imaging.

Inversion recovery sequences that vary the inversion time (t(i)) have been employed to determine T(1) and, more recently, quantitative magnetization transfer parameters. Specifically, in previous work, the inversion recovery pulse sequences varied t(i) only while maintaining a constant delay (t(d)) between repetitions. T(1) values were determined by fitting to a single exponential function, and quantitative magnetization transfer parameters were then determined by fitting to a biexponential function with an approximate solution. In the current study, new protocols are employed, which vary both t(i) and t(d) and fit the data with minimal approximations. Cramer-Rao lower bounds are calculated to search for acquisition schemes that will maximize the precision efficiencies of T(1) and quantitative magnetization transfer parameters. This approach is supported by Monte Carlo simulations. The optimal T(1) schemes are verified by measurements on MnCl(2) samples. The optimal quantitative magnetization transfer schemes are confirmed by measurements on a series of cross-linked bovine serum albumin phantoms of varying concentrations. The effects of varying the number of sampling data points are also explored, and a rapid acquisition scheme is demonstrated in vivo. These new optimized quantitative imaging methods provide an improved means for determining T(1) and magnetization transfer parameter values compared to previous inversion recovery based methods.