Search

This literature review presents a comprehensive overview of machine learning (ML) applications in proton MR spectroscopy (MRS). As the use of ML techniques in MRS continues to grow, this review aims to provide the MRS community with a structured overview of the state‐of‐the‐art methods. Specifically, we examine and summarize studies published between 2017 and 2023 from major journals in the MR field. We categorize these studies based on a typical MRS workflow, including data acquisition, processing, analysis, and artificial data generation. Our review reveals that ML in MRS is still in its early stages, with a primary focus on processing and analysis techniques, and less attention given to data acquisition. We also found that many studies use similar model architectures, with little comparison to alternative architectures. Additionally, the generation of artificial data is a crucial topic, with no consistent method for its generation. Furthermore, many studies demonstrate that artificial data suffers from generalization issues when tested on in vivo data. We also conclude that risks related to ML models should be addressed, particularly for clinical applications. Therefore, output uncertainty measures and model biases are critical to investigate. Nonetheless, the rapid development of ML in MRS and the promising results from the reviewed studies justify further research in this field. [ABSTRACT FROM AUTHOR]

Functional magnetic resonance imaging (fMRI) enables sites of brain activation to be localized in human subjects. For auditory system studies, however, the acoustic noise generated by the scanner tends to interfere with the assessments of this activation. Understanding and modeling fMRI acoustic noise is a useful step to its reduction. To study acoustic noise, the MR scanner is modeled as a linear electroacoustical system generating sound pressure signals proportional to the time derivative of the input gradient currents. The transfer function of one MR scanner is determined for two different input specifications: 1) by using the gradient waveform calculated by the scanner software and 2) by using a recording of the gradient current. Up to 4 kHz, the first method is shown as reliable as the second one, and its use is encouraged when direct measurements of gradient currents are not possible. Additionally, the linear order and average damping properties of the gradient coil system are determined by impulse response analysis. Since fMRI is often based on echo planar imaging (EPI) sequences, a useful validation of the transfer function prediction ability can be obtained by calculating the acoustic output for the EPI sequence. We found a predicted sound pressure level (SPL) for the EPI sequence of 104 dB SPL compared to a measured value of 102 dB SPL. As yet, the predicted EPI pressure waveform shows similarity as well as some differences with the directly measured EPI pressure waveform. Index Terms--Acoustic noise, fMRI, gradient noise, linear system, SPL.

The objective of the current study was to combine a time‐encoded pseudocontinuous arterial spin labeling (te‐pCASL) scheme with a golden angle radial readout for simultaneous acquisition of angiography and perfusion images from one single dataset, both in a highly flexible single‐slice approach as well as within a multislice setting. A te‐pCASL preparation and the golden angle radial readout were both used as a temporal resolution tool to retrospectively choose the temporal window for the reconstruction of both angiography and perfusion images from a single‐slice dataset. The temporal window could be chosen retrospectively and adjusted to the hemodynamics of the volunteer on the scanner for the single‐slice dataset. Angiographic images were reconstructed at a minimum temporal resolution of 69 ms. For the perfusion phase, only the densely sampled center of k‐space was included in the reconstruction. For a multislice acquisition, the golden angle radial readout allowed reconstruction of images with different spatial resolutions to provide angiographic and perfusion information over 10 slices. The te‐pCASL preparation was used as the only source for dynamic information. The multislice acquisition shows the ability of the golden angle radial readout to display the inflow of the labeled blood into the arteries as well as the perfusion in the tissue with full brain coverage. By combining a te‐pCASL preparation with a golden angle radial readout, single‐slice high temporal resolution angiography and good quality perfusion images were reconstructed in a flexible manner from a single dataset. Optimizing the golden angle radial readout for reconstructions at multiple spatial resolutions allows for multislice acquisition. [ABSTRACT FROM AUTHOR]

Purpose To compare methods for estimating B0 maps used in retrospective correction of high-resolution anatomical images at ultra-high field strength. The B0 maps were obtained using three methods: (1) 1D navigators and coil sensitivities, (2) field probe (FP) data and a low-order spherical harmonics model, and (3) FP data and a training-based model. Methods Data from nine subjects were acquired while they performed activities inducing B0 field fluctuations. Estimated B0 fields were compared with reference data, and the reductions of artifacts were compared in corrected [ABSTRACT FROM AUTHOR]

In this study, we present a new three-dimensional (3D), diffusion-prepared turbo spin echo sequence based on a stimulated-echo read-out (DPsti-TSE) enabling high-resolution and undistorted diffusion-weighted imaging (DWI). A dephasing gradient in the diffusion preparation module and rephasing gradients in the turbo spin echo module create stimulated echoes, which prevent signal loss caused by eddy currents. Near to perfect agreement of apparent diffusion coefficient (ADC) values between DPsti-TSE and diffusion-weighted echo planar imaging (DW-EPI) was demonstrated in both phantom transient signal experiments and phantom imaging experiments. High-resolution and undistorted DPsti-TSE was demonstrated in vivo in prostate and carotid vessel wall. 3D whole-prostate DWI was achieved with four b values in only 6 min. Undistorted ADC maps of the prostate peripheral zone were obtained at low and high imaging resolutions with no change in mean ADC values [(1.60 ± 0.10) × 10−3 versus (1.60 ± 0.02) × 10−3 mm2/s]. High-resolution 3D DWI of the carotid vessel wall was achieved in 12 min, with consistent ADC values [(1.40 ± 0.23) × 10−3 mm2/s] across different subjects, as well as slice locations through the imaging volume. This study shows that DPsti-TSE can serve as a robust 3D diffusion-weighted sequence and is an attractive alternative to the traditional two-dimensional DW-EPI approaches. [ABSTRACT FROM AUTHOR]

Purpose To explore the effects of high permittivity dielectric pads on the transmit and receive characteristics of a 3 Tesla body coil centered at the thighs, and their implications on image uniformity in receive array applications. Theory and Methods Transmit and receive profiles of the body coil with and without dielectric pads were simulated and measured in healthy volunteers. Parallel imaging was performed using sensitivity encoding (SENSE) with and without pads. An intensity correction filter was constructed from the measured receive profile of the body coil. Results Measured and simulated data show that the dielectric pads improve the transmit homogeneity of the body coil in the thighs, but decrease its receive homogeneity, which propagates into reconstruction algorithms in which the body coil is used as a reference. However, by correcting for the body coil reception profile this effect can be mitigated. Conclusion Combining high permittivity dielectric pads with an appropriate body coil receive sensitivity filter improves the image uniformity substantially compared with the situation without pads. Magn Reson Med 76:1951-1956, 2016. © 2015 International Society for Magnetic Resonance in Medicine [ABSTRACT FROM AUTHOR]

Studies have suggested that, in subjects with subjective cognitive impairment (SCI), Alzheimer's disease (AD)-like changes may occur in the brain. Recently, an in vivo study has indicated the potential of ultra-high-field MRI to visualize amyloid-beta (Aβ)-associated changes in the cortex in patients with AD, manifested by a phase shift on T2*-weighted MRI scans. The main aim of this study was to investigate whether cortical phase shifts on T2*-weighted images at 7 T in subjects with SCI can be detected, possibly implicating the deposition of Aβ plaques and associated iron. Cognitive tests and T2*-weighted scans using a 7-T MRI system were performed in 28 patients with AD, 18 subjects with SCI and 27 healthy controls (HCs). Cortical phase shifts were measured. Univariate general linear modeling and linear regression analysis were used to assess the association between diagnosis and cortical phase shift, and between cortical phase shift and the different neuropsychological tests, adjusted for age and gender. The phase shift (mean, 1.19; range, 1.00-1.35) of the entire cortex in AD was higher than in both SCI (mean, 0.85; range, 0.73-0.99; p < 0.001) and HC (mean, 0.94; range, 0.79-1.10; p < 0.001). No AD-like changes, e.g. increased cortical phase shifts, were found in subjects with SCI compared with HCs. In SCI, a significant association was found between memory function (Wechsler Memory Scale, WMS) and cortical phase shift ( β = -0.544, p = 0.007). The major finding of this study is that, in subjects with SCI, an increased cortical phase shift measured at high field is associated with a poorer memory performance, although, as a group, subjects with SCI do not show an increased phase shift compared with HCs. This increased cortical phase shift related to memory performance may contribute to the understanding of SCI as it is still unclear whether SCI is a sign of pre-clinical AD. Copyright © 2014 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Purpose Fluctuations of the background magnetic field (B0) due to body and breathing motion can lead to significant artifacts in brain imaging at ultrahigh field. Corrections based on real-time sensing using external field probes show great potential. This study evaluates different aspects of field interpolation from these probes into the brain which is implicit in such methods. Measurements and simulations were performed to quantify how well B0-fluctuations in the brain due to body and breathing motion are reflected in external field probe measurements. Methods Field probe measurements were compared with scanner acquired B0-maps from experiments with breathing and shoulder movements. A realistic simulation of B0-fluctuations caused by breathing was performed, and used for testing different sets of field probe positions. Results The B0-fluctuations were well reflected in the field probe measurements in the shoulder experiments, while the breathing experiments showed only moderate correspondence. The simulations showed the importance of the probe positions, and that performing full 3rd order corrections based on 16 field probes is not recommended. Conclusion Methods for quantitative assessment of the field interpolation problem were developed and demonstrated. Field corrections based on external field measurements show great potential, although potential pitfalls were identified. Magn Reson Med 75:2020-2030, 2016. © 2015 Wiley Periodicals, Inc. [ABSTRACT FROM AUTHOR]

The aim of this study was to implement and evaluate a flexible and time-efficient interleaved imaging approach for the acquisition of proton and sodium images of the human knee at 7 T within a clinically relevant timescale. A flexible software framework was established which allowed the interleaving of multiple, different, fully specific absorption ratio (SAR)-validated scans. The system was able to switch between these different scans at flexible time points. The practical example presented consists of interleaved proton (Dixon imaging and T2* mapping) and sodium (mapping the sodium content and fluid-suppressed component separately) sequences with the key idea to perform proton MRI whilst the sodium nuclei relax towards thermal equilibrium, and vice versa. Comparisons were made between these four scans being acquired sequentially in the normal mode of scanner operation and those acquired in an interleaved fashion. Images acquired in the interleaved mode were very similar to those acquired in sequential scans with no image artifacts produced by the slight intra-sequence variation in steady-state magnetization. A reduction in scanning time of almost a factor of two was established using the interleaved scans, allowing such a protocol to be completed within 30 min. Phantom experiments and in vivo scans performed in healthy volunteers and in one patient proved the basic feasibility of this approach. This approach for the interleaving of multiple proton and sodium scans, each with different contrasts, is an efficient method for the design of new practical clinical protocols for sodium MRI. Copyright © 2015 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Purpose High field T2*-weighted MR images of the cerebral cortex are increasingly used to study tissue susceptibility changes related to aging or pathologies. This paper presents a novel automated method for the computation of quantitative cortical measures and group-wise comparison using 7 Tesla T2*-weighted magnitude and phase images. Methods The cerebral cortex was segmented using a combination of T2*-weighted magnitude and phase information and subsequently was parcellated based on an anatomical atlas. Local gray matter (GM)/white matter (WM) contrast and cortical profiles, which depict the magnitude or phase variation across the cortex, were computed from the magnitude and phase images in each parcellated region and further used for group-wise comparison. Differences in local GM/WM contrast were assessed using linear regression analysis. Regional cortical profiles were compared both globally and locally using permutation testing. The method was applied to compare a group of 10 young volunteers with a group of 15 older subjects. Results Using local GM/WM contrast, significant differences were revealed in at least 13 of 17 studied regions. Highly significant differences between cortical profiles were shown in all regions. Conclusion The proposed method can be a useful tool for studying cortical changes in normal aging and potentially in neurodegenerative diseases. Magn Reson Med 74:240-248, 2015. © 2014 Wiley Periodicals, Inc. [ABSTRACT FROM AUTHOR]

In the evaluation of cerebrovascular CO2 reactivity measurements, it is often assumed that the diameter of the large intracranial arteries insonated by transcranial Doppler remains unaffected by changes in arterial CO2 partial pressure. However, the strong cerebral vasodilatory capacity of CO2 challenges this assumption, suggesting that there should be some changes in diameter, even if very small. Data from previous studies on effects of CO2 on cerebral artery diameter [middle cerebral artery (MCA)] have been inconsistent. In this study, we examined 10 healthy subjects (5 women, 5 men, age 21–30 yr). High-resolution (0.2 mm in-plane) MRI scans at 7 Tesla were used for direct observation of the MCA diameter during hypocapnia, -1 kPa (-7.5 mmHg), normocapnia, 0 kPa (0 mmHg), and two levels of hypercapnia, +1 and +2 kPa (7.5 and 15 mmHg), with respect to baseline. The vessel lumen was manually delineated by two independent observers. The results showed that the MCA diameter increased by 6.8 ± 2.9% in response to 2 kPa end-tidal PCO2 (PETCO2) above baseline. However, no significant changes in diameter were observed at the -1 kPa (-1.2 ± 2.4%), and +1 kPa 9+1.4 ± 3.2%) levels relative to normocapnia. The nonlinear response of the MCA diameter to CO2 was fitted as a continuous calibration curve. Cerebral blood flow changes measured by transcranial Doppler could be corrected by this calibration curve using concomitant PETCO2 measurements. In conclusion, the MCA diameter remains constant during small deviations of the PETCO2 from normocapnia, but increases at higher PETCO2 values. [ABSTRACT FROM AUTHOR]

Purpose To tailor and optimize the Dual Refocusing Echo Acquisition Mode (DREAM) approach for volumetric B1+ mapping of the brain at 7T. Theory and Methods A new DREAM echo timing scheme based on the virtual stimulated echo was derived to minimize potential effects of transverse relaxation. Furthermore, the DREAM B1+ mapping performance was investigated in simulations and experimentally in phantoms and volunteers for volumetric applications, studying and optimizing the accuracy of the sequence with respect to saturation effects, slice profile imperfections, and T1 and T2 relaxation. Volumetric brain protocols were compiled for different isotropic resolutions (5-2.5 mm) and SENSE factors, and were studied in vivo for different RF drive modes (circular/linear polarization) and the application of dielectric pads. Results Volumetric B1+ maps with good SNR at 2.5 mm isotropic resolution were acquired in about 20 s or less. The specific absorption rate was well below the safety limits for all scans. Mild flow artefacts were observed in the large vessels. Moreover, a slight contrast in the ventricle was observed in the B1+ maps, which could be attributed to T1 and T2 relaxation effects. Conclusion DREAM enables safe, very fast, and robust volumetric B1+ mapping of the brain at ultrahigh fields. Magn Reson Med 71:246-256, 2014. © 2013 Wiley Periodicals, Inc. [ABSTRACT FROM AUTHOR]

Purpose: To propose a new method that integrates both magnitude and phase information obtained from magnetic resonance (MR) T*2-weighted scans for cerebral cortex segmentation of the elderly. Materials and Methods: This method makes use of K-means clustering on magnitude and phase images to compute an initial segmentation, which is further refined by means of transformation with reconstruction criteria. The method was evaluated against the manual segmentation of 7T in vivo MR data of 20 elderly subjects (age = 67.7 ± 10.9). The added value of combining magnitude and phase was also evaluated by comparing the performance of the proposed method with the results obtained when limiting the available data to either magnitude or phase. Results: The proposed method shows good overlap agreement, as quantified by the Dice Index (0.79 ± 0.04), limited bias (average relative volume difference = 2.94%), and reasonable volumetric correlation ( R = 0.555, p = 0.011). Using the combined magnitude and phase information significantly improves the segmentation accuracy compared with using either magnitude or phase. Conclusion: This study suggests that the proposed method is an accurate and robust approach for cerebral cortex segmentation in datasets presenting low gray/white matter contrast. J. Magn. Reson. Imaging 2012;36:99-109. © 2012 Wiley Periodicals Inc. [ABSTRACT FROM AUTHOR]

Large variations of tip angle within a slice can lead to suboptimal pulse power optimization using standard techniques, which measure the average tip angle over a slice; this effect is especially pronounced at fields of 7 T and above. A technique was introduced that performed a volume-selective power optimization in less than 10 sec and automatically calibrates the radiofrequency pulses for subsequent spectroscopy scans. Using this technique, MR spectra were acquired in the human calf of seven healthy volunteers with a partial volume Tx/Rx coil. Increases in signal-to-noise ratio based upon the unsuppressed water signal between 22 ± 5% and 166 ± 42%, compared to spectra obtained with the conventional power calibration technique, were measured in different regions of the calf muscle. This new technique was able to measure the inhomogeneous radiofrequency field at 7 T and its use resulted in a considerable signal-to-noise ratio increase. Magn Reson Med, 2010. © 2009 Wiley-Liss, Inc. [ABSTRACT FROM AUTHOR]

Functional magnetic resonance imaging (fMRI) enables sites of brain activation to be localized in human subjects. For auditory system studies, however, the acoustic noise generated by the scanner tends to interfere with the assessments of this activation. Understanding and modeling fMRI acoustic noise is a useful step to its reduction. To study acoustic noise, the MR scanner is modeled as a linear electroacoustical system generating sound pressure signals proportional to the time derivative of the input gradient currents. The transfer function of one MR scanner is determined for two different input specifications: 1) by using the gradient waveform calculated by the scanner software and 2) by using a recording of the gradient current. Up to 4 kHz, the first method is shown as reliable as the second one, and its use is encouraged when direct measurements of gradient currents are not possible. Additionally, the linear order and average damping properties of the gradient coil system are determined by impulse response analysis. Since fMRI is often based on echo planar imaging (EPI) sequences, a useful validation of the transfer function prediction ability can be obtained by calculating the acoustic output for the EPI sequence. We found a predicted sound pressure level (SPL) for the EPI sequence of 104 dB SPL compared to a measured value of 102 dB SPL. As yet, the predicted EPI pressure waveform shows similarity as well as some differences with the directly measured EPI pressure waveform. [ABSTRACT FROM AUTHOR]

Functional MRI (fMRI) generally employs gradient-echo echo-planar imaging (GE-EPI) to measure blood oxygen level-dependent (BOLD) signal changes that result from changes in tissue relaxation time T [ABSTRACT FROM AUTHOR]

In this article, the authors focus on study done on the wall of basilar artery with the help of Seven Tesla Magnetic Resonance Imaging (7T MRI) to diagnose perimesencephalic hemorrhage (PMH) in patients. The study states that the MRI and the computed tomography (CT) pattern showed the absence of abnormalities and intramural hematoma. The study mentions that pattern of CT disproves the hypothesis that PMH is caused by basilar artery wall abnormalities.