Time-domain semi-parametric estimation based on a metabolite basis set

Development of a 31P magnetic resonance spectroscopy technique to quantify NADH and NAD+ at 3 T

NADH and NAD+ act as electron donors and acceptors and NAD+ was shown to stimulate mitochondrial biogenesis and metabolic health. We here develop a non-invasive Phosphorous Magnetic Resonance Spectroscopy (31P-MRS) method to quantify these metabolites in human skeletal muscle on a clinical 3 T MRI scanner. This new MR-sequence enables NADH and NAD+ quantification by suppressing α-ATP signal, normally overlapping with NADH and NAD+. The sequence is based on a double spin echo in combination with a modified z-Filter achieving strong α-ATP suppression with little effect on NAD+ and NADH. Here we test and validate it in phantoms and in humans by measuring reproducibility and detecting a physiological decrease in NAD+ and increase in NADH induced by ischemia. Furthermore, the 31P-MRS outcomes are compared to analysis in biopsies. Additionally, we show higher NAD+ and lower NADH content in physically active older adults compared to sedentary individuals, reflecting increased metabolic health.

TensorFit: A torch-based tool for ultrafast metabolite fitting of large MRSI data sets

PURPOSE

To introduce a tool (TensorFit) for ultrafast and robust metabolite fitting of MRSI data based on Torch's auto-differentiation and optimization framework.

METHODS

TensorFit was implemented in Python based on Torch's auto-differentiation to fit individual metabolites in MRS spectra. The underlying time domain and/or frequency domain fitting model is based on a linear combination of metabolite spectroscopic response. The computational time efficiency and accuracy of TensorFit were tested on simulated and in vivo MRS data and compared against TDFDFit and QUEST.

RESULTS

TensorFit demonstrates a significant improvement in computation speed, achieving a 165-times acceleration compared with TDFDFit and 115 times against QUEST. TensorFit showed smaller percentual errors on simulated data compared with TDFDFit and QUEST. When tested on in vivo data, it performed similarly to TDFDFit with a 2% better fit in terms of mean squared error while obtaining a 169-fold speedup.

CONCLUSION

TensorFit enables fast and robust metabolite fitting in large MRSI data sets compared with conventional metabolite fitting methods. This tool could boost the clinical applicability of large 3D MRSI by enabling the fitting of large MRSI data sets within computation times acceptable in a clinical environment.

Mapping intracellular NAD content in entire human brain using phosphorus-31 MR spectroscopic imaging at 7 Tesla

Introduction

Nicotinamide adenine dinucleotide (NAD) is a crucial molecule in cellular metabolism and signaling. Mapping intracellular NAD content of human brain has long been of interest. However, the sub-millimolar level of cerebral NAD concentration poses significant challenges for in vivo measurement and imaging.

Methods

In this study, we demonstrated the feasibility of non-invasively mapping NAD contents in entire human brain by employing a phosphorus-31 magnetic resonance spectroscopic imaging (31P-MRSI)-based NAD assay at ultrahigh field (7 Tesla), in combination with a probabilistic subspace-based processing method.

Results

The processing method achieved about a 10-fold reduction in noise over raw measurements, resulting in remarkably reduced estimation errors of NAD. Quantified NAD levels, observed at approximately 0.4 mM, exhibited good reproducibility within repeated scans on the same subject and good consistency across subjects in group data (2.3 cc nominal resolution). One set of higher-resolution data (1.0 cc nominal resolution) unveiled potential for assessing tissue metabolic heterogeneity, showing similar NAD distributions in white and gray matter. Preliminary analysis of age dependence suggested that the NAD level decreases with age.

Discussion

These results illustrate favorable outcomes of our first attempt to use ultrahigh field 31P-MRSI and advanced processing techniques to generate a whole-brain map of low-concentration intracellular NAD content in the human brain.

Magnetic Resonance Spectroscopy Quantification Aided by Deep Estimations of Imperfection Factors and Macromolecular Signal

OBJECTIVE

Magnetic Resonance Spectroscopy (MRS) is an important technique for biomedical detection. However, it is challenging to accurately quantify metabolites with proton MRS due to serious overlaps of metabolite signals, imperfections because of non-ideal acquisition conditions, and interference with strong background signals mainly from macromolecules. The most popular method, LCModel, adopts complicated non-linear least square to quantify metabolites and addresses these problems by designing empirical priors such as basis-sets, imperfection factors. However, when the signal-to-noise ratio of MRS signal is low, the solution may have large deviation.

METHODS

Linear Least Squares (LLS) is integrated with deep learning to reduce the complexity of solving this overall quantification. First, a neural network is designed to explicitly predict the imperfection factors and the overall signal from macromolecules. Then, metabolite quantification is solved analytically with the introduced LLS. In our Quantification Network (QNet), LLS takes part in the backpropagation of network training, which allows the feedback of the quantification error into metabolite spectrum estimation. This scheme greatly improves the generalization to metabolite concentrations unseen in training compared to the end-to-end deep learning method.

RESULTS

Experiments show that compared with LCModel, the proposed QNet, has smaller quantification errors for simulated data, and presents more stable quantification for 20 healthy in vivo data at a wide range of signal-to-noise ratio. QNet also outperforms other end-to-end deep learning methods.

CONCLUSION

This study provides an intelligent, reliable and robust MRS quantification.

SIGNIFICANCE

QNet is the first LLS quantification aided by deep learning.

CloudBrain-MRS: An intelligent cloud computing platform for in vivo magnetic resonance spectroscopy preprocessing, quantification, and analysis

Magnetic resonance spectroscopy (MRS) is an important clinical imaging method for diagnosis of diseases. MRS spectrum is used to observe the signal intensity of metabolites or further infer their concentrations. Although the magnetic resonance vendors commonly provide basic functions of spectrum plots and metabolite quantification, the spread of clinical research of MRS is still limited due to the lack of easy-to-use processing software or platform. To address this issue, we have developed CloudBrain-MRS, a cloud-based online platform that provides powerful hardware and advanced algorithms. The platform can be accessed simply through a web browser, without the need of any program installation on the user side. CloudBrain-MRS also integrates the classic LCModel and advanced artificial intelligence algorithms and supports batch preprocessing, quantification, and analysis of MRS data from different vendors. Additionally, the platform offers useful functions: (1) Automatically statistical analysis to find biomarkers for diseases; (2) Consistency verification between the classic and artificial intelligence quantification algorithms; (3) Colorful three-dimensional visualization for easy observation of individual metabolite spectrum. Last, data of both healthy subjects and patients with mild cognitive impairment are used to demonstrate the functions of the platform. To the best of our knowledge, this is the first cloud computing platform for in vivo MRS with artificial intelligence processing. We have shared our cloud platform at MRSHub, providing at least two years of free access and service. If you are interested, please visit https://mrshub.org/software_all/#CloudBrain-MRS or https://csrc.xmu.edu.cn/CloudBrain.html.

Alteration of skeletal muscle energy metabolism assessed by phosphorus-31 magnetic resonance spectroscopy in clinical routine, part 1: Advanced quality control pipeline

Implementing a standardized phosphorus-31 magnetic resonance spectroscopy (31 P-MRS) dynamic acquisition protocol to evaluate skeletal muscle energy metabolism and monitor muscle fatigability, while being compatible with various longitudinal clinical studies on diversified patient cohorts, requires a high level of technicality and expertise. Furthermore, processing data to obtain reliable results also demands a great degree of expertise from the operator. In this two-part article, we present an advanced quality control approach for data acquired using a dynamic 31 P-MRS protocol. The aim is to provide decision support to the operator to assist in data processing and obtain reliable results based on objective criteria. We present here, in part 1, an advanced data quality control (QC) approach of a dynamic 31 P-MRS protocol. Part 2 is an impact study that will demonstrate the added value of the QC approach to explore data derived from two clinical populations that experience significant fatigue, patients with coronavirus disease 2019 and multiple sclerosis. In part 1, 31 P-MRS was performed using 3-T clinical MRI in 175 subjects from clinical and healthy control populations conducted in a University Hospital. An advanced data QC score (QCS) was developed using multiple objective criteria. The criteria were based on current recommendations from the literature enriched by new proposals based on clinical experience. The QCS was designed to indicate valid and corrupt data and guide necessary objective data editing to extract as much valid physiological data as possible. Dynamic acquisitions using an MR-compatible ergometer ran over a rest (40 s), exercise (2 min), and a recovery phase (6 min). Using QCS enabled rapid identification of subjects with data anomalies, allowing the user to correct the data series or reject them partially or entirely, as well as identify fully valid datasets. Overall, the use of the QCS resulted in the automatic classification of 45% of the subjects, including 58 participants who had data with no criterion violation and 21 participants with violations that resulted in the rejection of all dynamic data. The remaining datasets were inspected manually with guidance, allowing acceptance of full datasets from an additional 80 participants and recovery phase data from an additional 16 subjects. Overall, more anomalies occurred with patient data (35% of datasets) compared with healthy controls (15% of datasets). In conclusion, the QCS ensures a standardized data rejection procedure and rigorous objective analysis of dynamic 31 P-MRS data obtained from patients. This methodology contributes to efforts made to standardize 31 P-MRS practices that have been underway for a decade, with the goal of making it an empowered tool for clinical research.

Alteration of skeletal muscle energy metabolism assessed by 31 P MRS in clinical routine: Part 2. Clinical application

In this second part of a two-part paper, we intend to demonstrate the impact of the previously proposed advanced quality control pipeline. To understand its benefit and challenge the proposed methodology in a real scenario, we chose to compare the outcome when applying it to the analysis of two patient populations with significant but highly different types of fatigue: COVID-19 and multiple sclerosis (MS). 31 P-MRS was performed on a 3 T clinical MRI, in 19 COVID-19 patients, 38 MS patients, and 40 matched healthy controls. Dynamic acquisitions using an MR-compatible ergometer ran over a rest (40 s), exercise (2 min), and a recovery phase (6 min). Long and short TR acquisitions were also made at rest for T1 correction. The advanced data quality control pipeline presented in Part 1 is applied to the selected patient cohorts to investigate its impact on clinical outcomes. We first used power and sample size analysis to estimate objectively the impact of adding the quality control score (QCS). Then, comparisons between patients and healthy control groups using the validated QCS were performed using unpaired t tests or Mann-Whitney tests (p < 0.05). The application of the QCS resulted in increased statistical power, changed the values of several outcome measures, and reduced variability (standard deviation). A significant difference was found between the T1PCr and T1Pi values of MS patients and healthy controls. Furthermore, the use of a fixed correction factor led to systematically higher estimated concentrations of PCr and Pi than when using individually corrected factors. We observed significant differences between the two patient populations and healthy controls for resting [PCr]-MS only, [Pi ], [ADP], [H2 PO4 - ], and pH-COVID-19 only, and post-exercise [PCr], [Pi ], and [H2 PO4 - ]-MS only. The dynamic indicators τPCr , τPi , ViPCr , and Vmax were reduced for COVID-19 and MS patients compared with controls. Our results show that QCS in dynamic 31 P-MRS studies results in smaller data variability and therefore impacts study sample size and power. Although QCS resulted in discarded data and therefore reduced the acceptable data and subject numbers, this rigorous and unbiased approach allowed for proper assessment of muscle metabolites and metabolism in patient populations. The outcomes include an increased metabolite T1 , which directly affects the T1 correction factor applied to the amplitudes of the metabolite, and a prolonged τPCr , indicating reduced muscle oxidative capacity for patients with MS and COVID-19.

Quantification of short echo time MRS signals with improved version of QUantitation based on quantum ESTimation algorithm

Magnetic resonance spectroscopy offers information about metabolite changes in the organism, which can be used in diagnosis. While short echo time proton spectra exhibit more distinguishable metabolites compared with proton spectra acquired with long echo times, their quantification (and providing estimates of metabolite concentrations) is more challenging. They are hampered by a background signal, which originates mainly from macromolecules (MM) and mobile lipids. An improved version of the quantification algorithm QUantitation based on quantum ESTimation (QUEST), with MM prior knowledge (QUEST-MM), dedicated to proton signals and invoking appropriate prior knowledge on MM, is proposed and tested. From a single acquisition, it enables better metabolite quantification, automatic estimation of the background, and additional automatic quantification of MM components, thus improving its applicability in the clinic. The proposed algorithm may facilitate studies that involve patients with pathological MM in the brain. QUEST-MM and three QUEST-based strategies for quantifying short echo time signals are compared in terms of bias-variance trade-off and Cramér-Rao lower bound estimates. The performances of the methods are evaluated through extensive Monte Carlo studies. In particular, the histograms of the metabolite and MM amplitude distributions demonstrate the performances of the estimators. They showed that QUEST-MM works better than QUEST (Subtract approach) and is a good alternative to QUEST when measured MM signal is unavailable or unsuitable. Quantification with QUEST-MM is shown for 1 H in vivo rat brain signals obtained with the SPECIAL pulse sequence at 9.4 T, and human brain signals obtained, respectively, with STEAM at 4 T and PRESS at 3 T. QUEST-MM is implemented in jMRUI and will be available for public use from version 7.1.

A review of machine learning applications for the proton MR spectroscopy workflow

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.

Quantification of MR spectra by deep learning in an idealized setting: Investigation of forms of input, network architectures, optimization by ensembles of networks, and training bias

PURPOSE

The aims of this work are (1) to explore deep learning (DL) architectures, spectroscopic input types, and learning designs toward optimal quantification in MR spectroscopy of simulated pathological spectra; and (2) to demonstrate accuracy and precision of DL predictions in view of inherent bias toward the training distribution.

METHODS

Simulated 1D spectra and 2D spectrograms that mimic an extensive range of pathological in vivo conditions are used to train and test 24 different DL architectures. Active learning through altered training and testing data distributions is probed to optimize quantification performance. Ensembles of networks are explored to improve DL robustness and reduce the variance of estimates. A set of scores compares performances of DL predictions and traditional model fitting (MF).

RESULTS

Ensembles of heterogeneous networks that combine 1D frequency-domain and 2D time-frequency domain spectrograms as input perform best. Dataset augmentation with active learning can improve performance, but gains are limited. MF is more accurate, although DL appears to be more precise at low SNR. However, this overall improved precision originates from a strong bias for cases with high uncertainty toward the dataset the network has been trained with, tending toward its average value.

CONCLUSION

MF mostly performs better compared to the faster DL approach. Potential intrinsic biases on training sets are dangerous in a clinical context that requires the algorithm to be unbiased to outliers (i.e., pathological data). Active learning and ensemble of networks are good strategies to improve prediction performances. However, data quality (sufficient SNR) has proven as a bottleneck for adequate unbiased performance-like in the case of MF.

Physics-informed deep learning approach to quantification of human brain metabolites from magnetic resonance spectroscopy data

PURPOSE

While the recommended analysis method for magnetic resonance spectroscopy data is linear combination model (LCM) fitting, the supervised deep learning (DL) approach for quantification of MR spectroscopy (MRS) and MR spectroscopic imaging (MRSI) data recently showed encouraging results; however, supervised learning requires ground truth fitted spectra, which is not practical. Moreover, this work investigates the feasibility and efficiency of the LCM-based self-supervised DL method for the analysis of MRS data.

METHOD

We present a novel DL-based method for the quantification of relative metabolite concentrations, using quantum-mechanics simulated metabolite responses and neural networks. We trained, validated, and evaluated the proposed networks with simulated and publicly accessible in-vivo human brain MRS data and compared the performance with traditional methods. A novel adaptive macromolecule fitting algorithm is included. We investigated the performance of the proposed methods in a Monte Carlo (MC) study.

RESULT

The validation using low-SNR simulated data demonstrated that the proposed methods could perform quantification comparably to other methods. The applicability of the proposed method for the quantification of in-vivo MRS data was demonstrated. Our proposed networks have the potential to reduce computation time significantly.

CONCLUSION

The proposed model-constrained deep neural networks trained in a self-supervised manner can offer fast and efficient quantification of MRS and MRSI data. Our proposed method has the potential to facilitate clinical practice by enabling faster processing of large datasets such as high-resolution MRSI datasets, which may have thousands of spectra.

Joint spectral quantification of MR spectroscopic imaging using linear tangent space alignment-based manifold learning

PURPOSE

To develop a manifold learning-based method that leverages the intrinsic low-dimensional structure of MR Spectroscopic Imaging (MRSI) signals for joint spectral quantification.

METHODS

A linear tangent space alignment (LTSA) model was proposed to represent MRSI signals. In the proposed model, the signals of each metabolite were represented using a subspace model and the local coordinates of the subspaces were aligned to the global coordinates of the underlying low-dimensional manifold via linear transform. With the basis functions of the subspaces predetermined via quantum mechanics simulations, the global coordinates and the matrices for the local-to-global coordinate alignment were estimated by fitting the proposed LTSA model to noisy MRSI data with a spatial smoothness constraint on the global coordinates and a sparsity constraint on the matrices.

RESULTS

The performance of the proposed method was validated using numerical simulation data and in vivo proton-MRSI experimental data acquired on healthy volunteers at 3T. The results of the proposed method were compared with the QUEST method and the subspace-based method. In all the compared cases, the proposed method achieved superior performance over the QUEST and the subspace-based methods both qualitatively in terms of noise and artifacts in the estimated metabolite concentration maps, and quantitatively in terms of spectral quantification accuracy measured by normalized root mean square errors.

CONCLUSION

Joint spectral quantification using linear tangent space alignment-based manifold learning improves the accuracy of MRSI spectral quantification.

Experimental Basis Sets of Quantification of Brain 1H-Magnetic Resonance Spectroscopy at 3.0 T

In vivo short echo time (TE) proton magnetic resonance spectroscopy (1H-MRS) is a useful method for the quantification of human brain metabolites. The purpose of this study was to evaluate the performance of an in-house, experimentally measured basis set and compare it with the performance of a vendor-provided basis set. A 3T clinical scanner with 32-channel receive-only phased array head coil was used to generate 16 brain metabolites for the metabolite basis set. For voxel localization, point-resolved spin-echo sequence (PRESS) was used with volume of interest (VOI) positioned at the center of the phantoms. Two different basis sets were subjected to linear combination of model spectra of metabolite solutions in vitro (LCModel) analysis to evaluate the in-house acquired in vivo 1H-MR spectra from the left prefrontal cortex of 22 healthy subjects. To evaluate the performance of the two basis sets, the Cramer-Rao lower bounds (CRLBs) of each basis set were compared. The LCModel quantified the following metabolites and macromolecules: alanine (Ala), aspartate (Asp), γ-amino butyric acid (GABA), glucose (Glc), glutamine (Gln), glutamate (Glu), glutathione (GHS), Ins (myo-Inositol), lactate (Lac), N-acetylaspartate (NAA), N-acetylaspartylglutamate (NAAG), taurine (Tau), phosphoryl-choline + glycerol-phosphoryl-choline (tCho), N-acetylaspartate + N-acetylaspartylglutamate (tNA), creatine + phosphocreatine (tCr), Glu + Gln (Glx) and Lip13a, Lip13b, Lip09, MM09, Lip20, MM20, MM12, MM14, MM17, Lip13a + Lip13b, MM14 + Lip13a + Lip13b + MM12, MM09 + Lip09, MM20 + Lip20. Statistical analysis showed significantly different CRLBs: Asp, GABA, Gln, GSH, Ins, Lac, NAA, NAAG, Tau, tCho, tNA, Glx, MM20, MM20 + Lip20 (p < 0.001), tCr, MM12, MM17 (p < 0.01), and Lip20 (p < 0.05). The estimated ratio of cerebrospinal fluid (CSF) in the region of interest was calculated to be about 5%. Fitting performances are better, for the most part, with the in-house basis set, which is more precise than the vendor-provided basis set. In particular, Asp is expected to have reliable CRLB (<30%) at high field (e.g., 3T) in the left prefrontal cortex of human brain. The quantification of Asp was difficult, due to the inaccuracy of Asp fitting with the vendor-provided basis set.

Metabolic changes in glioblastomas in response to choline kinase inhibition: In vivo MRS in rodent models

Changes in glioblastoma (GBM) metabolism was investigated in response to JAS239, a choline kinase inhibitor, using MRS. In addition to the inhibition of phosphocholine synthesis, we investigated changes in other key metabolic pathways associated with GBM progression and treatment response. Three syngeneic rodent models of GBM were used: F98 (N = 12) and 9L (N = 8) models in rats and GL261 (N = 10) in mice. Rodents were intracranially injected with GBM cells in the right cortex and tumor growth was monitored using T₂ -weighted images. Animals were treated once daily with intraperitoneal injections of 4 mg/kg JAS239 (F98 rats, n = 6; 9L rats, n = 6; GL261 mice, n = 5) or saline (control group, F98 rats, n = 6; 9L rats, n = 2; GL261 mice, n = 5) for five consecutive days. Single voxel spectra were acquired on Days 0 (T0, baseline) and 6 (T6, end of treatment) from the tumor as well as the contralateral normal brain using a PRESS sequence. Changes in metabolite ratios (tCho/tCr, tCho/NAA, mI/tCr, Glx/tCr and (Lip + Lac)/Cr) were used to assess metabolic pathway alterations in response to JAS239. Tumor growth arrest was noted in all models in response to JAS239 treatment compared with saline-treated animals, with a significant reduction (p < 0.05) in the F98 model. A reduction in tCho/tCr was observed with JAS239 treatment in all GBM models, indicating reduced phospholipid metabolism, with the highest reduction in 9L followed by GL261 and F98 tumors. A significant reduction (p < 0.05) in the tCho/NAA ratio was observed in the 9L model. A significant reduction in mI/tCr (p < 0.05) was found in JAS239-treated F98 tumors compared with the saline-treated animals. A non-significant trend of reduction in Glx/tCr was observed only in F98 and 9L tumors. JAS239-treated F98 tumors also showed a significant increase in Lip + Lac (p < 0.05), indicating increased cell death. This study demonstrated the utility of MRS in assessing metabolic changes in GBM in response to choline kinase inhibition.

Minority Stress and the Effects on Emotion Processing in Transgender Men and Cisgender People: A Study Combining fMRI and 1H-MRS

BACKGROUND

Minority stress via discrimination, stigmatization, and exposure to violence can lead to development of mood and anxiety disorders and underlying neurobiochemical changes. To date, the neural and neurochemical correlates of emotion processing in transgender people (and their interaction) are unknown.

METHODS

This study combined functional magnetic resonance imaging and magnetic resonance spectroscopy to uncover the effects of anxiety and perceived stress on the neural and neurochemical substrates, specifically choline, on emotion processing in transgender men. Thirty transgender men (TM), 30 cisgender men, and 35 cisgender women passively viewed angry, neutral, happy, and surprised faces in the functional magnetic resonance imaging scanner, underwent a magnetic resonance spectroscopy scan, and filled out mood- and anxiety-related questionnaires.

RESULTS

As predicted, choline levels modulated the relationship between anxiety and stress symptoms and the neural response to angry and surprised (but not happy faces) in the amygdala. This was the case only for TM but not cisgender comparisons. More generally, neural responses in the left amygdala, left middle frontal gyrus, and medial frontal gyrus to emotional faces in TM resembled that of cisgender women.

CONCLUSIONS

These results provide first evidence, to our knowledge, of a critical interaction between levels of analysis and that choline may influence neural processing of emotion in individuals prone to minority stress.

Respiratory-triggered quantitative MR spectroscopy of the human cervical spinal cord at 7 T

PURPOSE

Ultra-high field 1 H MR spectroscopy (MRS) is of great interest to help characterizing human spinal cord pathologies. However, very few studies have been reported so far in this small size structure at these fields due to challenging experimental difficulties caused by static and radiofrequency field heterogeneities, as well as physiological motion. In this work, in line with the recent developments proposed to strengthen spinal cord MRS feasibility at 7 T, a respiratory-triggered acquisition approach was optimized to compensate for dynamic B 0 field heterogeneities and to provide robust cervical spinal cord MRS data.

METHODS

A semi-LASER sequence was purposely used, and a dedicated raw data processing algorithm was developed to enhance MR spectral quality by discarding corrupted scans. To legitimate the choices done during the optimization stage, additional tests were carried out to determine the impact of breathing, voluntary motion, body mass index, and fitting algorithm. An in-house quantification tool was concomitantly designed for accurate estimation of the metabolite concentration ratios for choline, N-acetyl-aspartate (NAA), myo-inositol and glutathione. The method was tested on a cohort of 14 healthy volunteers.

RESULTS

Average water linewidth and NAA signal-to-noise ratio reached 0.04 ppm and 11.01, respectively. The group-average metabolic ratios were in good agreement with previous studies and showed intersession reproducibility variations below 30%.

CONCLUSION

The developed approach allows a rise of the acquired MRS signal quality and of the quantification robustness as compared to previous studies hence offering strengthened possibilities to probe the metabolism of degenerative and traumatic spinal cord pathologies.

A metabolic biomarker predicts Parkinson's disease at the early stages in patients and animal models

BackgroundCare management of Parkinson's disease (PD) patients currently remains symptomatic, mainly because diagnosis relying on the expression of the cardinal motor symptoms is made too late. Earlier detection of PD therefore represents a key step for developing therapies able to delay or slow down its progression.MethodsWe investigated metabolic markers in 3 different animal models of PD, mimicking different phases of the disease assessed by behavioral and histological evaluation, and in 3 cohorts of de novo PD patients and matched controls (n = 129). Serum and brain tissue samples were analyzed by nuclear magnetic resonance spectroscopy and data submitted to advanced multivariate statistics.ResultsOur translational strategy reveals common metabolic dysregulations in serum of the different animal models and PD patients. Some of them were mirrored in the tissue samples, possibly reflecting pathophysiological mechanisms associated with PD development. Interestingly, some metabolic dysregulations appeared before motor symptom emergence and could represent early biomarkers of PD. Finally, we built a composite biomarker with a combination of 6 metabolites. This biomarker discriminated animals mimicking PD from controls, even from the first, nonmotor signs and, very interestingly, also discriminated PD patients from healthy subjects.ConclusionFrom our translational study, which included 3 animal models and 3 de novo PD patient cohorts, we propose a promising biomarker exhibiting a high accuracy for de novo PD diagnosis that may possibly predict early PD development, before motor symptoms appear.FundingFrench National Research Agency (ANR), DOPALCOMP, Institut National de la Santé et de la Recherche Médicale, Université Grenoble Alpes, Association France Parkinson.

Neurochemical abnormalities in chronic fatigue syndrome: a pilot magnetic resonance spectroscopy study at 7 Tesla

RATIONALE

Chronic fatigue syndrome (CFS) is a common and burdensome illness with a poorly understood pathophysiology, though many of the characteristic symptoms are likely to be of brain origin. The use of high-field proton magnetic resonance spectroscopy (MRS) enables the detection of a range of brain neurochemicals relevant to aetiological processes that have been linked to CFS, for example, oxidative stress and mitochondrial dysfunction.

METHODS

We studied 22 CFS patients and 13 healthy controls who underwent MRS scanning at 7 T with a voxel placed in the anterior cingulate cortex. Neurometabolite concentrations were calculated using the unsuppressed water signal as a reference.

RESULTS

Compared to controls, CFS patients had lowered levels of glutathione, total creatine and myo-inositol in anterior cingulate cortex. However, when using N-acetylaspartate as a reference metabolite, only myo-inositol levels continued to be significantly lower in CFS participants.

CONCLUSIONS

The changes in glutathione and creatine are consistent with the presence of oxidative and energetic stress in CFS patients and are potentially remediable by nutritional intervention. A reduction in myo-inositol would be consistent with glial dysfunction. However, the relationship of the neurochemical abnormalities to the causation of CFS remains to be established, and the current findings require prospective replication in a larger sample.

MR-Spectroscopy of GABA and Glutamate/Glutamine Concentrations in Auditory Cortex in Clinical High-Risk for Psychosis Individuals

Psychosis involves changes in GABAergic and glutamatergic neurotransmission in auditory cortex that could be important for understanding sensory deficits and symptoms of psychosis. However, it is currently unclear whether such deficits are present in participants at clinical high-risk for psychosis (CHR-P) and whether they are associated with clinical outcomes. Magnetic Resonance Spectroscopy (MEGAPRESS, 1H-MRS at 3 Tesla) was used to estimate GABA, glutamate, and glutamate-plus-glutamine (Glx) levels in auditory cortex in a large sample of CHR-P (n = 99), CHR-N (clinical high-risk negative, n = 32), and 45 healthy controls. Examined were group differences in metabolite concentrations as well as relationships with clinical symptoms, general cognition, and 1-year follow-up clinical and general functioning in the CHR-P group. Results showed a marginal (p = 0.039) main group effect only for Glx, but not for GABA and glutamate concentrations, and only in left, not right, auditory cortex. This effect did not survive multiple comparison correction, however. Exploratory post-hoc tests revealed that there were significantly lower Glx levels (p = 0.029, uncorrected) in the CHR-P compared to the CHR-N group, but not relative to healthy controls (p = 0.058, uncorrected). Glx levels correlated with the severity of perceptual abnormalities and disorganized speech scores. However, in the CHR-P group, Glx levels did not predict clinical or functional outcomes. Accordingly, the findings from the present study suggest that MRS-measured GABA, glutamate and Glx levels in auditory cortex of CHR-P individuals are largely intact.

Results and interpretation of a fitting challenge for MR spectroscopy set up by the MRS study group of ISMRM

PURPOSE

Fitting of MRS data plays an important role in the quantification of metabolite concentrations. Many different spectral fitting packages are used by the MRS community. A fitting challenge was set up to allow comparison of fitting methods on the basis of performance and robustness.

METHODS

Synthetic data were generated for 28 datasets. Short-echo time PRESS spectra were simulated using ideal pulses for the common metabolites at mostly near-normal brain concentrations. Macromolecular contributions were also included. Modulations of signal-to-noise ratio (SNR); lineshape type and width; concentrations of γ-aminobutyric acid, glutathione, and macromolecules; and inclusion of artifacts and lipid signals to mimic tumor spectra were included as challenges to be coped with.

RESULTS

Twenty-six submissions were evaluated. Visually, most fit packages performed well with mostly noise-like residuals. However, striking differences in fit performance were found with bias problems also evident for well-known packages. In addition, often error bounds were not appropriately estimated and deduced confidence limits misleading. Soft constraints as used in LCModel were found to substantially influence the fitting results and their dependence on SNR.

CONCLUSIONS

Substantial differences were found for accuracy and precision of fit results obtained by the multiple fit packages.

Brain mitochondrial dysfunction and driving simulator performance in untreated obstructive sleep apnea

It is challenging to determine which patients with obstructive sleep apnea (OSA) have impaired driving ability. Vulnerability to this neurobehavioral impairment may be explained by lower brain metabolites levels involved in mitochondrial metabolism. This study compared markers of brain energy metabolism in OSA patients identified as vulnerable vs resistant to driving impairment following extended wakefulness. 44 patients with moderate-severe OSA underwent 28hr extended wakefulness with three 90min driving simulation assessments. Using a two-step cluster analysis, objective driving data (steering deviation and crashes) from the 2nd driving assessment (22.5 h awake) was used to categorise patients into vulnerable (poor driving, n = 21) or resistant groups (good driving, n = 23). ¹ H magnetic resonance spectra were acquired at baseline using two scan sequences (short echo PRESS and longer echo-time asymmetric PRESS), focusing on key metabolites, creatine, glutamate, N-acetylaspartate (NAA) in the hippocampus, anterior cingulate cortex and left orbito-frontal cortex. Based on cluster analysis, the vulnerable group had impaired driving performance compared with the resistant group and had lower levels of creatine (PRESS p = ns, APRESS p = 0.039), glutamate, (PRESS p < 0.01, APRESS p < 0.01), NAA (PRESS p = 0.038, APRESS p = 0.035) exclusively in the left orbito-frontal cortex. Adjusted analysis, higher glutamate was associated with a 21% (PRESS) and 36% (APRESS) reduced risk of vulnerable classification. Brain mitochondrial bioenergetics in the frontal brain regions are impaired in OSA patients who are vulnerable to driving impairment following sleep loss. These findings provide a potential way to identify at risk OSA phenotype when assessing fitness to drive, but this requires confirmation in larger future studies.

ProFit-1D-A 1D fitting software and open-source validation data sets

PURPOSE

Accurate and precise MRS fitting is crucial for metabolite concentration quantification of ¹ H-MRS spectra. LCModel, a spectral fitting software, has shown to have certain limitations to perform advanced spectral fitting by previous literature. Herein, we propose an open-source spectral fitting algorithm with adaptive spectral baseline determination and more complex cost functions.

THEORY

The MRS spectra are characterized by several parameters, which reflect the environment of the contributing metabolites, properties of the acquisition sequence, or additional disturbances. Fitting parameters should accurately describe these parameters. Baselines are also a major contributor to MRS spectra, in which smoothness of the spline baselines used for fitting can be adjusted based on the properties of the spectra. Three different cost functions used for the minimization problem were also investigated.

METHODS

The newly developed ProFit-1D fitting algorithm is systematically evaluated for simulations of several types of possible in vivo parameter variations. Although accuracy and precision are tested with simulated spectra, spectra measured in vivo at 9.4 T are used for testing precision using subsets of averages. ProFit-1D fitting results are also compared with LCModel.

RESULTS

Both ProFit-1D and LCModel fitted the spectra well with induced parameter and baseline variations. ProFit-1D proved to be more accurate than LCModel for simulated spectra. However, LCModel showed a somewhat increased precision for some spectral simulations and for in vivo data.

CONCLUSION

The open-source ProFit-1D fitting algorithm demonstrated high accuracy while maintaining precise metabolite concentration quantification. Finally, through the newly proposed cost functions, new ways to improve fitting were shown.

Characterization of the 1H-MRS Metabolite Spectra in Transgender Men with Gender Dysphoria and Cisgender People

Much research has been conducted on sexual differences of the human brain to determine whether and to what extent a brain gender exists. Consequently, a variety of studies using different neuroimaging techniques attempted to identify the existence of a brain phenotype in people with gender dysphoria (GD). However, to date, brain sexual differences at the metabolite level using magnetic resonance spectroscopy (1H-MRS) have not been explored in transgender people. In this study, 28 cisgender men (CM) and 34 cisgender women (CW) and 29 transgender men with GD (TMGD) underwent 1H-MRS at 3 Tesla MRI to characterize common brain metabolites. Specifically, levels of N-acetyl aspartate (NAA), choline (Cho), creatine (Cr), glutamate and glutamine (Glx), and myo-inositol + glycine (mI + Gly) were assessed in two brain regions, the amygdala-anterior hippocampus and the lateral parietal cortex. The results indicated a sex-assigned at birth pattern for Cho/Cr in the amygdala of TMGD. In the parietal cortex, a sex-assigned at birth and an intermediate pattern were found. Though assessed post-hoc, exploration of the age of onset of GD in TMGD demonstrated within-group differences in absolute NAA and relative Cho/Cr levels, suggestive for a possible developmental trend. While brain metabolite levels in TMGD resembled those of CW, some interesting findings, such as modulation of metabolite concentrations by age of onset of GD, warrant future inquiry.

Magnetic Resonance Spectroscopy of Hepatic Fat from Fundamental to Clinical Applications

The number of individuals suffering from fatty liver is increasing worldwide, leading to interest in the noninvasive study of liver fat. Magnetic resonance spectroscopy (MRS) is a powerful tool that allows direct quantification of metabolites in tissue or areas of interest. MRS has been applied in both research and clinical studies to assess liver fat noninvasively in vivo. MRS has also demonstrated excellent performance in liver fat assessment with high sensitivity and specificity compared to biopsy and other imaging modalities. Because of these qualities, MRS has been generally accepted as the reference standard for the noninvasive measurement of liver steatosis. MRS is an evolving technique with high potential as a diagnostic tool in the clinical setting. This review aims to provide a brief overview of the MRS principle for liver fat assessment and its application, and to summarize the current state of MRS study in comparison to other techniques.

Physical activity may contribute to brain health in multiple sclerosis: An MR volumetric and spectroscopy study

BACKGROUND AND PURPOSE

Physical activity may represent a disease-modifying therapy in persons with multiple sclerosis (pwMS). To date, there is limited research regarding mechanisms based on brain imaging for understanding the beneficial effects of physical activity in pwMS. This study examined the relationship between physical activity levels and thalamic and hippocampal volumes and brain metabolism in pwMS.

METHODS

The sample of 52 pwMS (37.3 ± 9.6 years of age; 35 females, 17 males) underwent a combination of volumetric magnetic resonance imaging and magnetic resonance spectroscopy. Current and lifetime physical activity were assessed using actigraphy and the adapted version of the Historical Activity Questionnaire, respectively.

RESULTS

Positive associations were observed between both actigraphy and self-reported levels of moderate-to-vigorous physical activity (MVPA) and thalamic and hippocampal volumes. Regarding brain metabolism, actigraphy and self-reported levels of MVPA were positively associated with higher hippocampal and thalamic levels of N-acetylaspartate/creatine ratio (NAA/Cr: marker of neural integrity and cell energy state).

CONCLUSIONS

This study provides novel evidence for a positive association between physical activity and thalamic and hippocampal volume and metabolism in pwMS. These findings support the hypothesis that physical activity, particularly MVPA, may serve as a disease-modifying treatment by improving brain health in pwMS.

Prodromal neuroinflammatory, cholinergic and metabolite dysfunction detected by PET and MRS in the TgF344-AD transgenic rat model of AD: a collaborative multi-modal study

Mouse models of Alzheimer's disease (AD) are valuable but do not fully recapitulate human AD pathology, such as spontaneous Tau fibril accumulation and neuronal loss, necessitating the development of new AD models. The transgenic (TG) TgF344-AD rat has been reported to develop age-dependent AD features including neuronal loss and neurofibrillary tangles, despite only expressing APP and PSEN1 mutations, suggesting an improved modelling of AD hallmarks. Alterations in neuronal networks as well as learning performance and cognition tasks have been reported in this model, but none have combined a longitudinal, multimodal approach across multiple centres, which mimics the approaches commonly taken in clinical studies. We therefore aimed to further characterise the progression of AD-like pathology and cognition in the TgF344-AD rat from young-adults (6 months (m)) to mid- (12 m) and advanced-stage (18 m, 25 m) of the disease. Methods: TgF344-AD rats and wild-type (WT) littermates were imaged at 6 m, 12 m and 18 m with [¹⁸F]DPA-714 (TSPO, neuroinflammation), [¹⁸F]Florbetaben (Aβ) and [¹⁸F]ASEM (α7-nicotinic acetylcholine receptor) and with magnetic resonance spectroscopy (MRS) and with (S)-[¹⁸F]THK5117 (Tau) at 15 and 25 m. Behaviour tests were also performed at 6 m, 12 m and 18 m. Immunohistochemistry (CD11b, GFAP, Aβ, NeuN, NeuroChrom) and Tau (S)-[¹⁸F]THK5117 autoradiography, immunohistochemistry and Western blot were also performed. Results: [¹⁸F]DPA-714 positron emission tomography (PET) showed an increase in neuroinflammation in TG vs wildtype animals from 12 m in the hippocampus (+11%), and at the advanced-stage AD in the hippocampus (+12%), the thalamus (+11%) and frontal cortex (+14%). This finding coincided with strong increases in brain microgliosis (CD11b) and astrogliosis (GFAP) at these time-points as assessed by immunohistochemistry. In vivo [¹⁸F]ASEM PET revealed an age-dependent increase uptake in the striatum and pallidum/nucleus basalis of Meynert in WT only, similar to that observed with this tracer in humans, resulting in TG being significantly lower than WT by 18 m. In vivo [¹⁸F]Florbetaben PET scanning detected Aβ accumulation at 18 m, and (S)-[¹⁸F]THK5117 PET revealed subsequent Tau accumulation at 25m in hippocampal and cortical regions. Aβ plaques were low but detectable by immunohistochemistry from 6 m, increasing further at 12 and 18 m with Tau-positive neurons adjacent to Aβ plaques at 18 m. NeuroChrom (a pan neuronal marker) immunohistochemistry revealed a loss of neuronal staining at the Aβ plaques locations, while NeuN labelling revealed an age-dependent decrease in hippocampal neuron number in both genotypes. Behavioural assessment using the novel object recognition task revealed that both WT & TgF344-AD animals discriminated the novel from familiar object at 3 m and 6 m of age. However, low levels of exploration observed in both genotypes at later time-points resulted in neither genotype successfully completing the task. Deficits in social interaction were only observed at 3 m in the TgF344-AD animals. By in vivo MRS, we showed a decrease in neuronal marker N-acetyl-aspartate in the hippocampus at 18 m (-18% vs age-matched WT, and -31% vs 6 m TG) and increased Taurine in the cortex of TG (+35% vs age-matched WT, and +55% vs 6 m TG). Conclusions: This multi-centre multi-modal study demonstrates, for the first time, alterations in brain metabolites, cholinergic receptors and neuroinflammation in vivo in this model, validated by robust ex vivo approaches. Our data confirm that, unlike mouse models, the TgF344-AD express Tau pathology that can be detected via PET, albeit later than by ex vivo techniques, and is a useful model to assess and longitudinally monitor early neurotransmission dysfunction and neuroinflammation in AD.

Contribution of macromolecules to brain 1 H MR spectra: Experts' consensus recommendations

Proton MR spectra of the brain, especially those measured at short and intermediate echo times, contain signals from mobile macromolecules (MM). A description of the main MM is provided in this consensus paper. These broad peaks of MM underlie the narrower peaks of metabolites and often complicate their quantification but they also may have potential importance as biomarkers in specific diseases. Thus, separation of broad MM signals from low molecular weight metabolites enables accurate determination of metabolite concentrations and is of primary interest in many studies. Other studies attempt to understand the origin of the MM spectrum, to decompose it into individual spectral regions or peaks and to use the components of the MM spectrum as markers of various physiological or pathological conditions in biomedical research or clinical practice. The aim of this consensus paper is to provide an overview and some recommendations on how to handle the MM signals in different types of studies together with a list of open issues in the field, which are all summarized at the end of the paper.

Preprocessing, analysis and quantification in single-voxel magnetic resonance spectroscopy: experts' consensus recommendations

Once an MRS dataset has been acquired, several important steps must be taken to obtain the desired metabolite concentration measures. First, the data must be preprocessed to prepare them for analysis. Next, the intensity of the metabolite signal(s) of interest must be estimated. Finally, the measured metabolite signal intensities must be converted into scaled concentration units employing a quantitative reference signal to allow meaningful interpretation. In this paper, we review these three main steps in the post-acquisition workflow of a single-voxel MRS experiment (preprocessing, analysis and quantification) and provide recommendations for best practices at each step.

Deep learning can accelerate and quantify simulated localized correlated spectroscopy

Nuclear magnetic resonance spectroscopy (MRS) allows for the determination of atomic structures and concentrations of different chemicals in a biochemical sample of interest. MRS is used in vivo clinically to aid in the diagnosis of several pathologies that affect metabolic pathways in the body. Typically, this experiment produces a one dimensional (1D) 1H spectrum containing several peaks that are well associated with biochemicals, or metabolites. However, since many of these peaks overlap, distinguishing chemicals with similar atomic structures becomes much more challenging. One technique capable of overcoming this issue is the localized correlated spectroscopy (L-COSY) experiment, which acquires a second spectral dimension and spreads overlapping signal across this second dimension. Unfortunately, the acquisition of a two dimensional (2D) spectroscopy experiment is extremely time consuming. Furthermore, quantitation of a 2D spectrum is more complex. Recently, artificial intelligence has emerged in the field of medicine as a powerful force capable of diagnosing disease, aiding in treatment, and even predicting treatment outcome. In this study, we utilize deep learning to: (1) accelerate the L-COSY experiment and (2) quantify L-COSY spectra. All training and testing samples were produced using simulated metabolite spectra for chemicals found in the human body. We demonstrate that our deep learning model greatly outperforms compressed sensing based reconstruction of L-COSY spectra at higher acceleration factors. Specifically, at four-fold acceleration, our method has less than 5% normalized mean squared error, whereas compressed sensing yields 20% normalized mean squared error. We also show that at low SNR (25% noise compared to maximum signal), our deep learning model has less than 8% normalized mean squared error for quantitation of L-COSY spectra. These pilot simulation results appear promising and may help improve the efficiency and accuracy of L-COSY experiments in the future.

Review and prospect: NMR spectroscopy denoising and reconstruction with low-rank Hankel matrices and tensors

Nuclear magnetic resonance (NMR) spectroscopy is an important analytical tool in chemistry, biology, and life science, but it suffers from relatively low sensitivity and long acquisition time. Thus, improving the apparent signal-to-noise ratio and accelerating data acquisition became indispensable. In this review, we summarize the recent progress on low-rank Hankel matrix and tensor methods, which exploit the exponential property of free-induction decay signals, to enable effective denoising and spectra reconstruction. We also outline future developments that are likely to make NMR spectroscopy a far more powerful technique.

In vivo γ-aminobutyric acid increase as a biomarker of the epileptogenic zone: An unbiased metabolomics approach

OBJECTIVE

Following surgery, focal seizures relapse in 20% to 50% of cases due to the difficulty of delimiting the epileptogenic zone (EZ) by current imaging or electrophysiological techniques. Here, we evaluate an unbiased metabolomics approach based on ex vivo and in vivo nuclear magnetic resonance spectroscopy (MRS) methods to discriminate the EZ in a mouse model of mesiotemporal lobe epilepsy (MTLE).

METHODS

Four weeks after unilateral injection of kainic acid (KA) into the dorsal hippocampus of mice (KA-MTLE model), we analyzed hippocampal and cortical samples with high-resolution magic angle spinning (HRMAS) magnetic resonance spectroscopy (MRS). Using advanced multivariate statistics, we identified the metabolites that best discriminate the injected dorsal hippocampus (EZ) and developed an in vivo MEGAPRESS MRS method to focus on the detection of these metabolites in the same mouse model.

RESULTS

Multivariate analysis of HRMAS data provided evidence that γ-aminobutyric acid (GABA) is largely increased in the EZ of KA-MTLE mice and is the metabolite that best discriminates the EZ when compared to sham and, more importantly, when compared to adjacent brain regions. These results were confirmed by capillary electrophoresis analysis and were not reversed by a chronic exposition to an antiepileptic drug (carbamazepine). Then, using in vivo noninvasive GABA-edited MRS, we confirmed that a high GABA increase is specific to the injected hippocampus of KA-MTLE mice.

SIGNIFICANCE

Our strategy using ex vivo MRS-based untargeted metabolomics to select the most discriminant metabolite(s), followed by in vivo MRS-based targeted metabolomics, is an unbiased approach to accurately define the EZ in a mouse model of focal epilepsy. Results suggest that GABA is a specific biomarker of the EZ in MTLE.

Effect of Long-Term Retention of Gadolinium on Metabolism of Deep Cerebellar Nuclei After Repeated Injections of Gadodiamide in Rats

OBJECTIVES

The aim of this study was to determine potential metabolism and histological modifications due to gadolinium retention within deep cerebellar nuclei (DCN) after linear gadolinium-based contrast agent injection (gadodiamide) in rats at 1 year after the last injection.

MATERIALS AND METHODS

Twenty female rats received 20 doses of gadodiamide (0.6 mmol of gadolinium per kilogram each) over 5 weeks. They were followed at 1 week (M0), 6 weeks (M1), and 54 to 55 weeks (M13) postinjections to evaluate hypersignal on unenhanced T1-weighted magnetic resonance imaging and metabolic alterations by H magnetic resonance spectroscopy (H-MRS). At 1 year postinjections, brains were sampled to determine the localization of gadolinium within cerebellum by laser ablation inductively coupled mass spectroscopy and to evaluate morphological changes by semiquantitative immunofluorescence analysis.

RESULTS

There is a significant increase of the ratio DCN/brainstem for the gadodiamide group at M0 (+7.2% vs control group = 0.989 ± 0.01), M1 (+7.6% vs control group = 1.002 ± 0.018), and it lasted up to M13 (+4.7% vs control group = 0.9862 ± 0.008). No variation among metabolic markers (cellular homeostasis [creatine, choline, taurine], excitatory neurotransmitter [glutamate], and metabolites specific to a cellular compartment [N-acetyl aspartate for neurons and myo-inositol for glial cells]) were detected by H-MRS between gadodiamide and saline groups at M0, M1, and M13. At M13, laser ablation inductively coupled mass spectroscopy demonstrated that long-term gadolinium retention occurred preferentially in DCN. No histological abnormalities (including analysis of astrocytes, neurons, and microglial cells) were found in the rostral part of DCN.

CONCLUSIONS

Repeated administration of gadodiamide lead to a retention of gadolinium preferentially within DCN at 1 year postinjections. This retention did not lead to any detectable changes of the measured metabolic biomarkers nor histological alterations.

Adaptive baseline fitting for 1 H MR spectroscopy analysis

PURPOSE

Accurate baseline modeling is essential for reliable MRS analysis and interpretation-particularly at short echo-times, where enhanced metabolite information coincides with elevated baseline interference. The degree of baseline smoothness is a key analysis parameter for metabolite estimation, and in this study, a new method is presented to estimate its optimal value.

METHODS

An adaptive baseline fitting algorithm (ABfit) is described, incorporating a spline basis into a frequency-domain analysis model, with a penalty parameter to enforce baseline smoothness. A series of candidate analyses are performed over a range of smoothness penalties, as part of a 4-stage algorithm, and the Akaike information criterion is used to estimate the appropriate penalty. ABfit is applied to a set of simulated spectra with differing baseline features and experimentally acquired 2D MRSI-both at a field strength of 3 Tesla.

RESULTS

Simulated analyses demonstrate metabolite errors result from 2 main sources: bias from an inflexible baseline (underfitting) and increased variance from an overly flexible baseline (overfitting). In the case of an ideal flat baseline, ABfit is shown to correctly estimate a highly rigid baseline, and for more realistic spectra a reasonable compromise between bias and variance is found. Analysis of experimentally acquired data demonstrates good agreement with known correlations between metabolite ratios and the contributing volumes of gray and white matter tissue.

CONCLUSIONS

ABfit has been shown to perform accurate baseline estimation and is suitable for fully automated routine MRS analysis.

Number of subjects required in common study designs for functional GABA magnetic resonance spectroscopy in the human brain at 3 Tesla

Magnetic resonance spectroscopy (MRS) is a research tool for measuring the concentration of metabolites such as γ-aminobutyric acid (GABA) and glutamate in the brain. MEGA-PRESS has been the preferred pulse sequence for GABA measurements due to low physiological GABA concentrations, hence low signal. To compensate, researchers incorporate long acquisition durations (7-10 min) making functional measurements of this metabolite challenging. Here, the acquisition duration and sample sizes required to detect specific concentration changes in GABA using MEGA-PRESS at 3 T are presented for both between-groups and within-session study designs. 75 spectra were acquired during rest using MEGA-PRESS from 41 healthy volunteers in 6 different brain regions at 3 T with voxel sizes between 13 and 22 cm3 . Between-group and within-session variance was calculated for different acquisition durations and power calculations were performed to determine the number of subjects required to detect a given percentage change in GABA/NAA signal ratio. Within-subject variability was assessed by sampling different segments of a single acquisition. Power calculations suggest that detecting a 15% change in GABA using a 2 min acquisition and a 27 cm3 voxel size, depending on the region, requires between 8 and 93 subjects using a within-session design. A between-group design typically requires more participants to detect the same difference. In brain regions with suboptimal shimming, the subject numbers can be up to 4-fold more. Collecting data for longer than 4 min in brain regions examined in this study is deemed unnecessary, as variance in the signal did not reduce further for longer durations.

CRPS Is Not Associated with Altered Sensorimotor Cortex GABA or Glutamate

Complex regional pain syndrome (CRPS) is a debilitating chronic pain disorder typically in the upper or lower limbs. While CRPS usually develops from a peripheral event, it is likely maintained by CNS changes. Indeed, CRPS is reported to be associated with sensorimotor cortex changes, or functional “reorganization,” as well as deficits such as poor tactile acuity. While the mechanisms underpinning cortical reorganization in CRPS are unknown, some have hypothesized that it involves disinhibition (i.e., a reduction in GABA activity). In this study, we addressed this hypothesis by using edited magnetic resonance spectroscopy to determine sensorimotor GABA and glutamate concentrations in 16 humans with CRPS and 30 matched control subjects and the relationship of these concentrations with tactile acuity. We found that individuals with upper limb CRPS displayed reduced tactile acuity in the painful hand, compared with the nonpainful hand and pain-free control subjects. Despite this acuity deficit, CRPS was not associated with altered GABA or glutamate concentrations within the sensorimotor cortex on either the side that represents the affected or unaffected hand. Furthermore, there was no significant relationship between sensorimotor GABA or glutamate concentrations and tactile acuity in CRPS subjects or control subjects. Although our sample was small, these data suggest that CRPS is not associated with altered total sensorimotor GABA or glutamate concentrations. While these results are at odds with the sensorimotor cortex disinhibition hypothesis, it is possible that GABAergic mechanisms other than total GABA concentration may contribute to such disinhibition.

Ultrafast magnetic resonance spectroscopic imaging using SPICE with learned subspaces

PURPOSE

To develop a subspace learning method for the recently proposed subspace-based MRSI approach known as SPICE, and achieve ultrafast 1 H-MRSI of the brain.

THEORY AND METHODS

A novel strategy is formulated to learn a low-dimensional subspace representation of MR spectra from specially acquired training data and use the learned subspace for general MRSI experiments. Specifically, the subspace learning problem is formulated as learning "empirical" distributions of molecule-specific spectral parameters (e.g., concentrations, lineshapes, and frequency shifts) by integrating physics-based model and the training data. The learned spectral parameters and quantum mechanical simulation basis can then be combined to construct acquisition-specific subspace for spatiospectral encoding and processing. High-resolution MRSI acquisitions combining ultrashort-TE/short-TR excitation, sparse sampling, and the elimination of water suppression have been performed to evaluate the feasibility of the proposed method.

RESULTS

The accuracy of the learned subspace and the capability of the proposed method in producing high-resolution 3D 1 H metabolite maps and high-quality spatially resolved spectra (with a nominal resolution of ∼2.4 × 2.4 × 3 mm3 in 5 minutes) were demonstrated using phantom and in vivo studies. By eliminating water suppression, we are also able to extract valuable information from the water signals for data processing ( B 0 map, frequency drift, and coil sensitivity) as well as for mapping tissue susceptibility and relaxation parameters.

CONCLUSIONS

The proposed method enables ultrafast 1 H-MRSI of the brain using a learned subspace, eliminating the need of acquiring subject-dependent navigator data (known as D 1 ) in the original SPICE technique. It represents a new way to perform MRSI experiments and an important step toward practical applications of high-resolution MRSI.

Improved Low-Rank Filtering of MR Spectroscopic Imaging Data With Pre-Learnt Subspace and Spatial Constraints

OBJECTIVE

To investigate the use of pre-learnt subspace and spatial constraints for denoising magnetic resonance spectroscopic imaging (MRSI) data.

METHOD

We exploit the partial separability or subspace structures of high-dimensional MRSI data for denoising. More specifically, we incorporate a subspace model with pre-learnt spectral basis into the low-rank approximation (LORA) method. Spectral basis is determined based on empirical prior distributions of the spectral parameters variations learnt from auxiliary training data; spatial priors are also incorporated as is done in LORA to further improve denoising performance.

RESULTS

The effects of the explicit subspace and spatial constraints in reducing estimation bias and variance have been analyzed using Cramér-Rao Lower bound analysis, Monte-Carlo study, and experimental study.

CONCLUSION

The denoising effectiveness of LORA can be significantly improved by incorporating pre-learnt spectral basis and spatial priors into LORA.

SIGNIFICANCE

This study provides an effective method for denoising MRSI data along with comprehensive analyses of its performance. The proposed method is expected to be useful for a wide range of studies using MRSI.

Analysis of metabolic abnormalities in high-grade glioma using MRSI and convex NMF

Clinical use of MRSI is limited by the level of experience required to properly translate MRSI examinations into relevant clinical information. To solve this, several methods have been proposed to automatically recognize a predefined set of reference metabolic patterns. Given the variety of metabolic patterns seen in glioma patients, the decision on the optimal number of patterns that need to be used to describe the data is not trivial. In this paper, we propose a novel framework to (1) separate healthy from abnormal metabolic patterns and (2) retrieve an optimal number of reference patterns describing the most important types of abnormality. Using 41 MRSI examinations (1.5 T, PRESS, T_E 135 ms) from 22 glioma patients, four different patterns describing different types of abnormality were detected: edema, healthy without Glx, active tumor and necrosis. The identified patterns were then evaluated on 17 MRSI examinations from nine different glioma patients. The results were compared against BraTumIA, an automatic segmentation method trained to identify different tumor compartments on structural MRI data. Finally, the ability to predict future contrast enhancement using the proposed approach was also evaluated.

Methodological consensus on clinical proton MRS of the brain: Review and recommendations

Proton MRS (1 H MRS) provides noninvasive, quantitative metabolite profiles of tissue and has been shown to aid the clinical management of several brain diseases. Although most modern clinical MR scanners support MRS capabilities, routine use is largely restricted to specialized centers with good access to MR research support. Widespread adoption has been slow for several reasons, and technical challenges toward obtaining reliable good-quality results have been identified as a contributing factor. Considerable progress has been made by the research community to address many of these challenges, and in this paper a consensus is presented on deficiencies in widely available MRS methodology and validated improvements that are currently in routine use at several clinical research institutions. In particular, the localization error for the PRESS localization sequence was found to be unacceptably high at 3 T, and use of the semi-adiabatic localization by adiabatic selective refocusing sequence is a recommended solution. Incorporation of simulated metabolite basis sets into analysis routines is recommended for reliably capturing the full spectral detail available from short TE acquisitions. In addition, the importance of achieving a highly homogenous static magnetic field (B0 ) in the acquisition region is emphasized, and the limitations of current methods and hardware are discussed. Most recommendations require only software improvements, greatly enhancing the capabilities of clinical MRS on existing hardware. Implementation of these recommendations should strengthen current clinical applications and advance progress toward developing and validating new MRS biomarkers for clinical use.

Fitting algorithms and baseline correction influence the results of non-invasive in vivo quantitation of 2-hydroxyglutarate with 1 H-MRS

¹ H-MRS enables non-invasive detection of 2-hydroxyglutarate (2-HG), an oncometabolite accumulating in gliomas carrying mutations in the isocitrate dehydrogenase (IDH) genes. Reliable 2-HG quantitation requires reproducible post-processing, deployment of fitting algorithms and quantitation methods. We prospectively enrolled 38 patients with suspected or recently diagnosed gliomas (IDH mutated n = 26). The MRI protocol included a ¹ H single voxel PRESS sequence with volumes of usually 8 mL or more (20 × 20 × 20 mm³ ) at T_E  = 97 ms and 180° pulse spacing. Our aim was to evaluate the reliability of 2-HG quantitation comparing two frequently used software tools and their respective options of baseline correction (jMRUI with the time domain methods AQSES and QUEST, and LCModel, which analyzes the frequency domain data). For AQSES, degrees of freedom for baseline correction constrains were varied. For LCModel, baseline correction was obtained with and without correction of the unknown background term (predefined macromolecules, lipids). Tissue concentrations were calculated based on the phantom replacement method. Quantitation of 2-HG levels showed similar mean 2-HG tissue concentrations for IDH mutated tumors (2.65mM, range 3.06-2.20) for all methods. Bland-Altman plots (difference plots) did not reveal a systematic bias (fixed bias) for any of the algorithms tested, and we were able to show a significant correlation regarding 2-HG concentration at the same echo time with few statistical outliers (parametric correlation). However, evaluation of outliers suggested that in vivo quantitation of 2-HG is affected not only by the fitting domain (time or frequency), but also by the baseline correction, which is a major contributing factor to the result of 2-HG fitting. Clinical application of 2-HG quantitation as a prognostic or predictive biomarker, particularly in multicenter trials, requires standardized use of fitting methods and baseline correction procedures.

Hypertonic sodium lactate reverses brain oxygenation and metabolism dysfunction after traumatic brain injury

BACKGROUND

The mechanisms by which hypertonic sodium lactate (HSL) solution act in injured brain are unclear. We investigated the effects of HSL on brain metabolism, oxygenation, and perfusion in a rodent model of diffuse traumatic brain injury (TBI).

METHODS

Thirty minutes after trauma, anaesthetised adult rats were randomly assigned to receive a 3 h infusion of either a saline solution (TBI-saline group) or HSL (TBI-HSL group). The sham-saline and sham-HSL groups received no insult. Three series of experiments were conducted up to 4 h after TBI (or equivalent) to investigate: 1) brain oedema using diffusion-weighted magnetic resonance imaging and brain metabolism using localized ¹H-magnetic resonance spectroscopy (n = 10 rats per group). The respiratory control ratio was then determined using oxygraphic analysis of extracted mitochondria, 2) brain oxygenation and perfusion using quantitative blood-oxygenation-level-dependent magnetic resonance approach (n = 10 rats per group), and 3) mitochondrial ultrastructural changes (n = 1 rat per group).

RESULTS

Compared with the TBI-saline group, the TBI-HSL and the sham-operated groups had reduced brain oedema. Concomitantly, the TBI-HSL group had lower intracellular lactate/creatine ratio [0.049 (0.047-0.098) vs 0.097 (0.079-0.157); P < 0.05], higher mitochondrial respiratory control ratio, higher tissue oxygen saturation [77% (71-79) vs 66% (55-73); P < 0.05], and reduced mitochondrial cristae thickness in astrocytes [27.5 (22.5-38.4) nm vs 38.4 (31.0-47.5) nm; P < 0.01] compared with the TBI-saline group. Serum sodium and lactate concentrations and serum osmolality were higher in the TBI-HSL than in the TBI-saline group.

CONCLUSIONS

These findings indicate that the hypertonic sodium lactate solution can reverse brain oxygenation and metabolism dysfunction after traumatic brain injury through vasodilatory, mitochondrial, and anti-oedema effects.

The relationship between thalamic GABA content and resting cortical rhythm in neuropathic pain

Recurrent thalamocortical connections are integral to the generation of brain rhythms and it is thought that the inhibitory action of the thalamic reticular nucleus is critical in setting these rhythms. Our work and others' has suggested that chronic pain that develops following nerve injury, that is, neuropathic pain, results from altered thalamocortical rhythm, although whether this dysrhythmia is associated with thalamic inhibitory function remains unknown. In this investigation, we used electroencephalography and magnetic resonance spectroscopy to investigate cortical power and thalamic GABAergic concentration in 20 patients with neuropathic pain and 20 pain-free controls. First, we found thalamocortical dysrhythmia in chronic orofacial neuropathic pain; patients displayed greater power than controls over the 4-25 Hz frequency range, most marked in the theta and low alpha bands. Furthermore, sensorimotor cortex displayed a strong positive correlation between cortical power and pain intensity. Interestingly, we found no difference in thalamic GABA concentration between pain subjects and control subjects. However, we demonstrated significant linear relationships between thalamic GABA concentration and enhanced cortical power in pain subjects but not controls. Whilst the difference in relationship between thalamic GABA concentration and resting brain rhythm between chronic pain and control subjects does not prove a cause and effect link, it is consistent with a role for thalamic inhibitory neurotransmitter release, possibly from the thalamic reticular nucleus, in altered brain rhythms in individuals with chronic neuropathic pain.

Quantification of GABA, glutamate and glutamine in a single measurement at 3 T using GABA-edited MEGA-PRESS

γ-Aminobutyric acid (GABA) and glutamate (Glu), major neurotransmitters in the brain, are recycled through glutamine (Gln). All three metabolites can be measured by magnetic resonance spectroscopy in vivo, although GABA measurement at 3 T requires an extra editing acquisition, such as Mescher-Garwood point-resolved spectroscopy (MEGA-PRESS). In a GABA-edited MEGA-PRESS spectrum, Glu and Gln co-edit with GABA, providing the possibility to measure all three in one acquisition. In this study, we investigated the reliability of the composite Glu + Gln (Glx) peak estimation and the possibility of Glu and Gln separation in GABA-edited MEGA-PRESS spectra. The data acquired in vivo were used to develop a quality assessment framework which identified MEGA-PRESS spectra in which Glu and Gln could be estimated reliably. Phantoms containing Glu, Gln, GABA and N-acetylaspartate (NAA) at different concentrations were scanned using GABA-edited MEGA-PRESS at 3 T. Fifty-six sets of spectra in five brain regions were acquired from 36 healthy volunteers. Based on the Glu/Gln ratio, data were classified as either within or outside the physiological range. A peak-by-peak quality assessment was performed on all data to investigate whether quality metrics can discriminate between these two classes of spectra. The quality metrics were as follows: the GABA signal-to-noise ratio, the NAA linewidth and the Glx Cramer-Rao lower bound (CRLB). The Glu and Gln concentrations were estimated with precision across all phantoms with a linear relationship between the measured and true concentrations: R¹ = 0.95 for Glu and R¹ = 0.91 for Gln. A quality assessment framework was set based on the criteria necessary for a good GABA-edited MEGA-PRESS spectrum. Simultaneous criteria of NAA linewidth <8 Hz and Glx CRLB <16% were defined as optimum features for reliable Glu and Gln quantification. Glu and Gln can be reliably quantified from GABA-edited MEGA-PRESS acquisitions. However, this reliability should be controlled using the quality assessment methods suggested in this work.

Longitudinal investigation of neuroinflammation and metabolite profiles in the APPswe ×PS1Δe9 transgenic mouse model of Alzheimer's disease

There is increasing evidence linking neuroinflammation to many neurological disorders including Alzheimer's disease (AD); however, its exact contribution to disease manifestation and/or progression is poorly understood. Therefore, there is a need to investigate neuroinflammation in both health and disease. Here, we investigate cognitive decline, neuroinflammatory and other pathophysiological changes in the APP_swe×PS1_Δe9 transgenic mouse model of AD. Transgenic (TG) mice were compared to C57BL/6 wild type (WT) mice at 6, 12 and 18 months of age. Neuroinflammation was investigated by [¹⁸F]DPA‐714 positron emission tomography and myo‐inositol levels using ¹H magnetic resonance spectroscopy (MRS) in vivo. Neuronal and cellular dysfunction was investigated by looking at N‐acetylaspartate (NAA), choline‐containing compounds, taurine and glutamate also using MRS. Cognitive decline was first observed at 12 m of age in the TG mice as assessed by working memory tests . A significant increase in [¹⁸F]DPA‐714 uptake was seen in the hippocampus and cortex of 18 m‐old TG mice when compared to age‐matched WT mice and 6 m‐old TG mice. No overall effect of gene was seen on metabolite levels; however, a significant reduction in NAA was observed in 18 m‐old TG mice when compared to WT. In addition, age resulted in a decrease in glutamate and an increase in choline levels. Therefore, we can conclude that increased neuroinflammation and cognitive decline are observed in TG animals, whereas NAA alterations occurring with age are exacerbated in the TG mice. These results support the role of neuroinflammation and metabolite alteration in AD and in ageing.

Mitochondrial dysfunction in myotonic dystrophy type 1

The pathophysiological mechanism linking the nucleotide expansion in the DMPK gene to the clinical manifestations of myotonic dystrophy type 1 (DM1) is still unclear. In vitro studies demonstrate DMPK involvement in the redox homeostasis of cells and the mitochondrial dysfunction in DM1, but in vivo investigations of oxidative metabolism in skeletal muscle have provided ambiguous results and have never been performed in the brain. Twenty-five DM1 patients (14M, 39 ± 11years) underwent brain proton MR spectroscopy (¹H-MRS), and sixteen cases (9M, 40 ± 13 years old) also calf muscle phosphorus MRS (³¹P-MRS). Findings were compared to those of sex- and age-matched controls. Eight DM1 patients showed pathological increase of brain lactate and, compared to those without, had larger lateral ventricles (p < 0.01), smaller gray matter volumes (p < 0.05) and higher white matter lesion load (p < 0.05). A reduction of phosphocreatine/inorganic phosphate (p < 0.001) at rest and, at first minute of exercise, a lower [phosphocreatine] (p = 0.003) and greater [ADP] (p = 0.004) were found in DM1 patients compared to controls. The post-exercise indices of muscle oxidative metabolism were all impaired in DM1, including the increase of time constant of phosphocreatine resynthesis (TC PCr, p = 0.038) and the reduction of the maximum rate of mitochondrial ATP synthesis (p = 0.033). TC PCr values correlated with the myotonic area score (ρ = 0.74, p = 0.01) indicating higher impairment of muscle oxidative metabolism in clinically more affected patients. Our findings provide clear in vivo evidence of multisystem impairment of oxidative metabolism in DM1 patients, providing a rationale for targeted treatment enhancing energy metabolism.