Vascular contributions to pattern analysis: Comparing gradient and spin echo fMRI at 3T

Alterations in density and morphology of the cerebral microvasculature have been reported to occur in Alzheimer's disease patients and animal models of the disease. In this study we compared magnetic resonance imaging (MRI) techniques for their utility to detect age-dependent changes of the cerebral vasculature in the arcAβ mouse model of cerebral amyloidosis. Dynamic susceptibility contrast (DSC)-MRI was performed by tracking the passage of a superparamagnetic iron oxide nanoparticle in the brain with dynamic gradient echo planar imaging (EPI). From this measurements relative cerebral blood volume [rCBV(DSC)] and relative cerebral blood flow (rCBF) were estimated. For the same animal maps of the relaxation shift index Q were computed from high resolution gradient echo and spin echo data that were acquired before and after superparamagnetic iron oxide (SPIO) nanoparticle injection. Q-values were used to derive estimates of microvessel density. The change in the relaxation rates obtained from pre- and post-contrast gradient echo data was used for the alternative determination of rCBV [rCBV()]. Linear mixed effects modeling found no significant association between rCBV(DSC), rCBV(), rCBF, and Q with genotype in 13-month old mice [compared to age-matched non-transgenic littermates (NTLs)] for any of the evaluated brain regions. In 24-month old mice there was a significant association for rCBV(DSC) with genotype in the cerebral cortex, and for rCBV() in the cerebral cortex and cerebellum. For rCBF there was a significant association in the cerebellum but not in other brain regions. Q-values in the olfactory bulb, cerebral cortex, striatum, hippocampus, and cerebellum in 24-month old mice were significantly associated with genotype. In those regions Q-values were reduced between 11 and 26% in arcAβ mice compared to age-matched NTLs. Vessel staining with CD31 immunohistochemistry confirmed a reduction of microvessel density in the old arcAβ mice. We further demonstrated a region-specific association between parenchymal and vascular deposition of β-amyloid and decreased vascular density, without a correlation with the amount of Aβ deposition. We found that Q mapping was more suitable than the hemodynamic read-outs to detect amyloid-related degeneration of the cerebral microvasculature.

4896113 Use of repeated gradient echoes for noise reduction and improved NMR imaging

Narrow bandwidth magnetic resonance (MR) imaging allows an increase of signal-to-noise ratio (SNR) but causes increased chemical shift and motion-induced artifacts. To obtain MR images with SNR approximately equal to that obtained with narrow bandwidth but with less chemical shift and motion-induced artifact, we introduced triple readout gradient reversal centered around the spin echo. As a result, signals from two gradient echoes and a single spin echo can be collected and summed. Phantom, knee, shoulder, and abdominal MR images were obtained using a 1.5 T GE Signa System at sampling rates ranging from 10 to 60 kHz. Since the bandwidth per pixel was tripled, chemical shift misregistration was reduced by the same factor. The summation image of two gradient echoes and one spin echo had an SNR comparable with that of a single spin echo acquired within the same total sampling interval. Data acquisition at a high sampling ratio also minimizes the dispersion of T2* weighting among three echoes. In addition, summation of the three resulting images decreases motion artifact by effective averaging.

Motivation: Recently, microstructure-informed myelin mapping (MIMM) was proposed to estimate myelin content from multi gradient-echo (mGRE) data by modeling both magnitude and phase-derived susceptibility values. Although magnitude modeling incorporated the susceptibility anisotropy, susceptibility modeling ignored it. Goal(s): Here, MIMoSA is proposed to improve upon MIMM by including the susceptibility anisotropy both in magnitude and phase. Approach: Following MIMM, a dictionary of voxel signals associated with specific microstructures including myelin and iron was generated with a novel negative susceptibility estimation approach. Results: In vivo results of healthy subjects showed that iMIMM performs similar with MIMM without the need of an additional orientation prior. Impact: By accurate modeling of the effect of white matter anisotropic susceptibility on the full gradient-echo signal, myelin content can be mapped from routine gradient echo acquisitions.

The purpose of the study was to compare the artefacts produced by different hip prostheses on magnetic resonance imaging (MRI). An identical MRI protocol was used to perform a quali-quantitative in vitro evaluation of artefacts caused by different hip prosthetic materials at different field strengths: prosthesis number 1, composed of cobalt-chrome-molybdenum (head and stem); prosthesis number 2, composed of ceramic (head) and titanium (stem); prosthesis number 3, composed of cobalt-chrome (head) and titanium (stem). All prostheses were imaged with both a clinical 1 Tesla (T) (Signa Horizon, General Electrics) and 1.5 T (Achieva, Philips) MRI system, using spin echo (SE) and gradient echo (GRE) sequences: sagittal T1 SE, coronal T2 fast SE (FSE), axial T1 SE, axial T2 FSE, sagittal T2 GRE, axial T2* GRE, coronal T1 GRE, axial T1 GRE. The artefacts produced by each prosthesis were assessed in each sequence at the different field strengths, by measuring the two longest diameters of the artefact in each section and sequence and comparing them to the actual diameters so as to obtain a ratio expressing the effective degree of artefact. Cobalt-chrome produced the largest artefacts both in SE (1.73 at 1 T and 2.37 at 1.5 T) and GRE sequences (2.8 at 1 T and 3.06 at 1.5 T) followed by titanium (SE, 1.6 at 1 T, 2.13 at 1.5 T; GRE, 2 at 1 T, 2.94 at 1.5 T) and cobalt-chrome-molybdenum (SE, 1.51 at 1 T, 1.67 at 1.5 T; GRE, 2.13 at 1 T and 2.48 at 1.5 T); ceramic produced the smallest artefacts in all sequences (SE, 1.0 at 1 T and 1.18 at 1.5 T; GRE, 1.3 at 1 T and 1.22 at 1.5T). Increasing the magnetic field strength, titanium showed the greatest variations in artefact size, and ceramic the smallest ones. The composition of prosthetic implants is decisive in determining the quality of MR imaging.

ObjectivesTo evaluate sources of error in the Magnetic Resonance Imaging (MRI) measurement of percent fibroglandular tissue (%FGT) using two-point Dixon sequences for fat-water separation.MethodsTen female volunteers (median age: 31 yrs, range: 23–50 yrs) gave informed consent following Research Ethics Committee approval. Each volunteer was scanned twice following repositioning to enable an estimation of measurement repeatability from high-resolution gradient-echo (GRE) proton-density (PD)-weighted Dixon sequences. Differences in measures of %FGT attributable to resolution, T1 weighting and sequence type were assessed by comparison of this Dixon sequence with low-resolution GRE PD-weighted Dixon data, and against gradient-echo (GRE) or spin-echo (SE) based T1-weighted Dixon datasets, respectively.Results%FGT measurement from high-resolution PD-weighted Dixon sequences had a coefficient of repeatability of ±4.3%. There was no significant difference in %FGT between high-resolution and low-resolution PD-weighted data. Values of %FGT from GRE and SE T1-weighted data were strongly correlated with that derived from PD-weighted data (r = 0.995 and 0.96, respectively). However, both sequences exhibited higher mean %FGT by 2.9% (p < 0.0001) and 12.6% (p < 0.0001), respectively, in comparison with PD-weighted data; the increase in %FGT from the SE T1-weighted sequence was significantly larger at lower breast densities.ConclusionAlthough measurement of %FGT at low resolution is feasible, T1 weighting and sequence type impact on the accuracy of Dixon-based %FGT measurements; Dixon MRI protocols for %FGT measurement should be carefully considered, particularly for longitudinal or multi-centre studies.

Myelin water fraction (MWF) mapping based on data fitting of a 3-pool exponential model of multi-echo gradient echo (mGRE) data using MRI shows great promises for in vivo myelin quantification. However, this multi-exponential fitting is ill-conditioned because of the similar relaxation times and frequency shifts of the various compartments. Additionally, the bound water residing in the myelin sheath of white matter is expected to have a faster longitudinal magnetisation recovery than that of the free water in the intra-axonal and extra-axonal space. When the Ernst angle is used to achieve maximum SNR and improve fitting, this will introduce a T1-weighting effect to the derived MWF. In this study, we first demonstrate that diffusion-weighted imaging can be used to infer the compartmental signal properties using an analytical fibre model to achieve a robust MWF estimation. Second, we show that by incorporating a variable flip angle scheme to the mGRE acquisition with a multi-compartment relaxometry model, not only the MWF is corrected from the T1 dependency but also the fitting procedure becomes less ill-conditioned and more SNR efficient. Finally, we demonstrate these two approaches can be combined to allow higher spatial resolution MWF maps than what has been reported to date with robust MWF estimation on a small cohort.

Functional magnetic resonance imaging (fMRI) is a unique window to the brain, enabling scientists to follow changes in brain activity in response to hormones, ageing, environment, drugs of abuse and other stimuli. There are two features that make fMRI unique when compared with other imaging modalities used in behavioural neuroscience. First, it can be entirely noninvasive: each animal can serve as its own control over the natural course of its life, vital for following neuroadaptation and other developmental processes critical to understanding behaviour. Second, fMRI has the spatial and temporal resolution to observe patterns of neuronal activity across the entire brain in less than a minute. Although fMRI does not have the cellular spatial resolution of immunostaining, nor the millisecond temporal resolution of electrophysiology, synchronised changes in neuronal activity across multiple brain areas seen with functional MRI can be viewed as functional neuroanatomical circuits coordinating the thoughts, memories and emotions for particular behaviours. Thus, fMRI affords a systems approach to the study of the brain, complementing and building from other neurobiological techniques to understand how behaviour is organised across multiple brain regions. In this review, we present a general background to fMRI and the different imaging modalities that can be used in fMRI studies. Included are examples of the application of fMRI in behavioural neuroscience research, along with discussion of the advantages and disadvantages of this technology.

High-field gradient-echo (GE) BOLD fMRI enables very high resolution imaging, and has great potential for detailed investigations of brain function. However, as spatial resolution increases, confounds due to signal from non-capillary vessels increasingly impact the fidelity of GE BOLD fMRI signals. Here we report on an assessment of the microvascular weighting of the GE BOLD response across the cortical depth in human cortex using spin-echo fMRI which is thought to be dominated by microvasculature (albeit not completely). BOLD responses were measured with a hemodynamic impulse response (HRF) obtained from the spin-echo (SE) and gradient-echo (GE) BOLD contrast using very short stimuli (0.25 s) and a fast event-related functional paradigm. We show that the onset (∼1.25 s) and the rising slope of the GE and SE HRFs are strikingly similar for voxels in deep gray matter presumably containing the most metabolically demanding neurons (layers III–IV). This finding provides a strong indication that the onset of the GE HRF in deep gray matter is predominantly associated with microvasculature.

Iron-oxide nanoparticles (IONPs) have shown tremendous utility for enhancing image contrast and delivering targeted therapies. Quantification of IONPs has been demonstrated at low concentrations with gradient echo (GRE) and spin echo (SE), and at high concentrations with echoless sequences such as swept imaging with Fourier transform (SWIFT). This work examines the overlap of IONP quantification with GRE, SE, and SWIFT. The limit of quantification of GRE, SE, inversion-recovery GRE, and SWIFT sequences was assessed using IONPs at a concentration range of 0.02 to 89.29 mM suspended in 1% agarose. Empirically derived limits of quantification were compared with International Union of Pure and Applied Chemistry definitions. Both commercial and experimental IONPs were used. All three IONPs assessed demonstrated an overlap of concentration quantification with GRE, SE, and SWIFT sequences. The largest dynamic range observed was 0.004 to 35.7 mM with Feraheme. The metrics established allow upper and lower quantitative limitations to be estimated given the relaxivity characteristics of the IONP and the concentration range of the material to be assessed. The methods outlined in this paper are applicable to any pulse sequence, IONP formulation, and field strength. Magn Reson Med 79:1420-1428, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Brain activation of cortical and subcortical foci during a complex motor task as demonstrated with spin-echo (SE) and gradient-echo (GE) FMRI and PET were compared in order to determine how signal changes in each of these FMRI techniques reflect cerebral blood flow changes shown by PET. Five normal volunteers performed bilateral fingering tapping in a defined sequential pattern during functional imaging. BOLD FMRI was performed at 1.5T. Five 32 second activation periods were alternated with five 32 second rest periods. MR signal detection was done by performing a least-squares fit of the time series data to a boxcar function. PET CBF measurements were performed using intravenous H 2 15 O. 4 pairs of control and task scans were obtained. PET activation maps were generated by subtracting the averaged activation and control scans. The statistical significance of regions of interest in sensorimotor cortex, basal ganglia and cerebellum for all three techniques were evaluated within each subject. GE-FMRI and SE-FMRI showed large amplitude (p&lt;0.001) signal increases in regions of sensorimotor cortex, basal ganglia and cerebellum. PET also showed significant (p ⩽ 0.01) increases in CBF in sensorimotor cortex and cerebellum but less significant increases in CBF in the basal ganglia (p&lt;0.16). The regions of cortical activation shown by SE-FMRI were smaller in size than those shown by GE-FMRI and corresponded more closely to the areas of increased CBF shown by PET. Both GE-FMRI and SE-FMRI are more sensitive for activation by this complex motor task along the entire motor pathway compared to H 2 15 O PET. SE-FMRI localizes areas of cortical activation more focally and corresponds more closely to areas of increased CBF on PET than does GE-FMRI.

Susceptibility artefacts from dental materials may compromise MRI diagnosis. However, little is known regarding MRI artefacts of dental material samples with the clinical shapes used in dentistry. The present phantom study aims to clarify how pulse sequences and sequence parameters affect MRI artefacts caused by metal-ceramic restorations. A phantom consisting of nickel-chromium metal-ceramic restorations (i.e. dental crowns and fixed bridges) and cylindrical reference specimens immersed in agar gel was imaged in 1.5 and 3.0 T MRI scanners. Gradient echo (GRE), spin echo (SE) and ultrashort echo time (UTE) pulse sequences were used. The artefact area in each image was automatically calculated from the pixel values within a region of interest. Mean values for similar pulse sequences differing in one parameter at a time were compared. A comparison between mean artefact area at 1.5 and 3.0 T, and from GRE and SE was also carried out. In addition, a parametric correlation between echo time (TE) and artefact area was performed. A significant correlation was found between TE and artefact area in GRE images. Higher receiver bandwidth significantly reduced artefact area in SE images. UTE images yielded the smallest artefact area at 1.5 T. In addition, a significant difference in mean artefact area was found between images at 1.5 and 3.0 T field strengths (p = 0.028) and between images from GRE and SE pulse sequences (p = 0.005). It is possible to compensate the effect of higher field strength on MRI artefacts by setting optimized pulse sequences for scanning patients with metal-ceramic restorations.

Multivariate Pattern Analysis of Functional Magnetic Resonance Imaging Data Reveals Deficits in Distributed Representations in Schizophrenia

Emerging evidence suggests that diabetes mellitus (DM) is associated with iron and calcium metabolism. However, few studies have investigated the presence of DM in cognitively impaired patients and its effect on brain iron and calcium accumulation. Therefore, we assessed the effects of DM on cognitively impaired patients using quantitative susceptibility mapping (QSM). From June 2012 to Feb 2014, 92 eligible cognitively impaired patients underwent 3T magnetic resonance imaging (MRI). There were 46 patients with DM (DM+) and 46 aged matched patients without DM (DM-). QSM was obtained from gradient echo data and analyzed by drawing regions of interest around relevant anatomical structures. Clinical factors and vascular pathology were also evaluated. Measurement differences between DM+ and DM- patients were assessed by t tests. A multiple regression analysis was performed to identify independent predictors of magnetic susceptibility. DM+ patients showed lower susceptibility values, indicative of lower brain iron content, than DM- patients, which was significant in the hippocampus (4.80 ± 8.31 ppb versus 0.22 ± 10.60 ppb, p = 0.024) and pulvinar of the thalamus (36.30 ± 19.88 ppb versus 45.90 ± 20.02 ppb, p = 0.023). On multiple regression analysis, microbleed number was a predictor of susceptibility change in the hippocampus (F = 4.291, beta = 0.236, p = 0.042) and DM was a predictor of susceptibility change in the pulvinar of the thalamus (F = 4.900, beta = - 0.251, p = 0.030). In cognitively impaired patients, presence of DM was associated with lower susceptibility change in the pulvinar of the thalamus and hippocampus. This suggests that there may be region-specific alterations of calcium deposition in cognitively impaired subjects with DM.

Naturalistic stimuli like movies evoke complex perceptual processes, which are of great interest in the study of human cognition by functional MRI (fMRI). However, conventional fMRI analysis based on statistical parametric mapping (SPM) and the general linear model (GLM) is hampered by a lack of accurate parametric models of the BOLD response to complex stimuli. In this situation, statistical machine-learning methods, a.k.a. multivariate pattern analysis (MVPA), have received growing attention for their ability to generate stimulus response models in a data-driven fashion. However, machine-learning methods typically require large amounts of training data as well as computational resources. In the past, this has largely limited their application to fMRI experiments involving small sets of stimulus categories and small regions of interest in the brain. By contrast, the present study compares several classification algorithms known as Nearest Neighbor (NN), Gaussian Naïve Bayes (GNB), and (regularized) Linear Discriminant Analysis (LDA) in terms of their classification accuracy in discriminating the global fMRI response patterns evoked by a large number of naturalistic visual stimuli presented as a movie. Results show that LDA regularized by principal component analysis (PCA) achieved high classification accuracies, above 90% on average for single fMRI volumes acquired 2 s apart during a 300 s movie (chance level 0.7% = 2 s/300 s). The largest source of classification errors were autocorrelations in the BOLD signal compounded by the similarity of consecutive stimuli. All classifiers performed best when given input features from a large region of interest comprising around 25% of the voxels that responded significantly to the visual stimulus. Consistent with this, the most informative principal components represented widespread distributions of co-activated brain regions that were similar between subjects and may represent functional networks. In light of these results, the combination of naturalistic movie stimuli and classification analysis in fMRI experiments may prove to be a sensitive tool for the assessment of changes in natural cognitive processes under experimental manipulation.