To be updated...

To be updated...

To be updated...

To be updated...

Magnetic resonance spectroscopy (MRS) offers unparalleled insights into tissue metabolism and molecular composition. However, in vivo 1H-MRS and MRS imaging (MRSI) have intrinsic low signal-to-noise ratio (SNR) due to the low concentrations of metabolites (1-10 mM). Additionally, separation of different resonances is difficult due to strong spectral overlap between different metabolites, e.g., N-acetylaspartate (NAA) with N-acetylaspartylglutamate (NAAG), creatine (Cr) with phosphocreatine (PCr), phosphocholine (PCho) with glycerophosphocholine (GPC), glutamate (Glu) with glutamine (Gln), and γ-aminobutyric acid (GABA) with Cr, PCr, glutathione and macromolecules.

The SNR was shown to increase roughly with the main magnetic field by B01.65 or B01.94, while the spectral dispersion increases linearly with B0. Therefore, SNR and spectral overlap can be mitigated by increasing the B0 and reducing intra-voxel dispersion via an improved resolution in MRSI. However, increased main magnetic field results in increased specific absorption rate (SAR), increased B0 and B1 inhomogeneities, and a higher spectral bandwidth needed to cover the spectral range of interest. B0- and B1-inhomogeneities result in broader spectral peaks, and lower SNR, respectively.

Although working at ultra-high fields poses some challenges, the ultra-high-field MR systems (7 T and 9.4 T) have revolutionized our ability to probe human brain metabolism. In my talk, I will focus on presenting our findings from single-voxel MRS and 3D MRSI of the human brain conducted with the world’s first 10.5 T whole-body MR system. 

This talk will first summarize the benefits and challenges of UHF for magnetic resonance spectroscopy (MRS) research, focusing on factors that impact the successful execution of clinical research studies. I will then outline the approaches we took to streamline acquisition of high quality MRS data at 7T. Finally, I will provide examples of clinical research studies that utilized the 7T platform, including investigations of cerebral effects of diabetes and pilot clinical trials of antioxidant supplements in neurodegenerative diseases. 

To be updated...

Purpose: To investigate the Influence of macromolecular background on the evaluation of glutamate (Glu) transverserelaxation time (T2).

Methods: In-vivo 1H-MR spectra were acquired at 7T (Siemens Terra scanner) with 1Tx/32Rx head coil and using the STEAM (TM/TR = 28.82/5500 ms) sequence. Data were acquired at six different TEs (5.12, 19, 32.5, 46, 60, 90 ms: 64 averages per TE). Spectroscopy voxel was placed in the occipital lobe (24x18x18mm3) of 6 healthy volunteers (HV’s). Data were fitted in LCModel with three different strategies: A) LCModel simulated macromolecules (SIM_MM), B) measured macromolecules with unmatched and fixed TE = 8ms (FIX_MM), C) parametrized macromolecules (PAR_MM). Glu T2 was computed for each subject and fitting strategy by fitting the Glu signal decay as a function of TE in MATLB: SGlu(TE) = SGlu(0 ms) x exp(-TE/T2). Paired t-test was performed on the Glu T2 between each fitting strategies.

Results: Spectra exhibited high signal-to-noise ratio and narrow linewidth (TE= 5.12 ms: SNR = 201.02 ± 35, LW = 11.6 ± 0.76;TE = 90 ms: SNR = 85.13 ± 10.63, LW = 10.45 ± 0.62). Fits for Glu signal decays are plotted in. PAR_MM provided significantly higher Glu T2 vs SIM_MM (p = 0.0001) and vs FIX_MM (p = 0.03).

To be updated...

To be updated...

In nature, form dictates function. In the human brain, the network of physical connections—the connectome—plays a crucial role in constraining and shaping patterns of neural activity, creating correlations between distant brain regions. In this talk, I present both empirical and computational evidence demonstrating that the connectome exerts a causal influence on the brain's whole-brain correlation structure, or functional connectivity (FC). I review several methodological approaches for assessing the coupling (or decoupling) of FC to network wiring, including correlational techniques and biophysical modeling. The focus, however, is on a modeling approach that lies between these extremes: communication dynamics. This approach employs stylized network models to explore specific hypotheses about the brain’s interareal communication policies.

Purpose: Neurological post-acute sequelae of COVID-19 (neuroPASC) is characterized by neurological symptoms that persist after SARS-CoV-2 infection. Chronic inflammation is believed to be one of the causes underlying neuroPASC. We investigated the link between systemic inflammation, white matter integrity and white matter perfusion in neuroPASC.

Methods: Participants with neuroPASC and controls without symptomatic COVID-19 were enrolled at five US sites. MRI data were acquired on SIEMENS Prisma 3T scanners. Diffusion and pseudocontinuous arterial spin labeling (pCASL) acquisitions and processing were based on the HCP Lifespan protocol. Fractional anisotropy (FA), axial-, radial-, and mean diffusivity (AD, RD, and MD, respectively) maps were generated using DTIFIT (FSL). Cerebral blood flow (CBF) and arterial transit time (ATT) were calculated as previously. We used MIX (microstructure imaging of crossing) to estimate NODDI model parameters and generate orientation dispersion index (ODI), neurite density index (NDI), and free water fraction (IsoVF) maps. Mean regional diffusion and perfusion metrics were extracted in the native space, using an atlas-based segmentation (JHU white matter atlas). Concentration of the inflammatory marker high-sensitivity C-reactive protein (hsCRP; mg/L) was measured in the plasma. We used ANCOVA followed by pairwise tests to compare neuroPASC groups to controls, while adjusting for age and sex. To evaluate associations between diffusion and perfusion metrics we used Pearson partial correlations adjusted for age and sex; associations between imaging metrics and hsCRP were assessed using Spearman partial correlations. Cohen’s d effect sizes were computed using covariance-corrected residuals, where linear regression was used to adjust for effects of age and sex.

Results: In total, 102 participants with neuroPASC (age=48±15 (mean±SD), 75% females) were assessed ~2 years (IQR, 1.3-2.8) after infection and compared to 74 control participants (age=44±17, 63% females) with no known SARS-CoV-2 infection. In the neuroPASC group, 25% participants were hospitalized during their COVID-19 illness (non-hospitalized: n=77, age=45±14, 72% females; hospitalized: n=25, age=56±12, 80% females). Hospitalized participants had higher levels of hsCRP in the plasma compared to controls (p=0.03). Participants with neuroPASC had widespread lower AD, higher ODI and higher CBF compared to controls, with larger effect sizes in the hospitalized vs control comparison. Higher CBF was associated with lower AD and higher ODI. Lower AD, higher ODI and higher CBF were associated with higher concentration of hsCRP in the plasma.

Discussion: Higher hsCRP in hospitalized participants provides evidence of systemic inflammation, ~2 years after acute infection. AD and ODI are thought to be biomarkers of focal inflammation in the brain. The link between CBF and chronic inflammation is unclear, with some studies showing reduced CBF in chronic inflammation, while others show higher CBF in patients with multiple sclerosis and high inflammation. Our study suggests that SARS-CoV-2 infection may lead to persistent inflammatory changes in the brain, altering white matter integrity and cerebral blood flow, which may contribute to lingering neurological symptoms.

Mapping the human connectome across micro-, meso-, and macro-scales depends on MRI systems with both unprecedented gradient performance and carefully engineered hardware to maintain stability and safety. The Connectome 2.0 scanner, developed through the NIH BRAIN Initiative at Massachusetts General Hospital, is a head-only 3T MRI platform featuring 500 mT/m maximum gradient strength and 600 T/m/s slew rate per axis (Ramos-Llorden et al., Nat Biomed Eng, 2025). Its three-layer gradient coil integrates shoulder cut-outs and force-balanced windings, raising peripheral nerve stimulation thresholds by ~2.4- to 4.2-fold over the original Connectome MRI scanner. Custom-engineered high-channel RF arrays, including a 72-channel in vivo head coil and 64-channel ex vivo brain coil, together with integrated field probes, enable precise trajectory monitoring and echo-time reduction, yielding SNR gains of at least 30% greater than the original Connectome MRI scanner and up to an order of magnitude over standard clinical scanners at high b-values. These capabilities open new territory for mesoscale connectomics. Ultra-strong gradients allow diffusion encoding at very short effective times, boosting sensitivity to micron-scale restrictions and supporting advanced models such as SANDI for soma and neurite density, laminar profiling of cortical architecture, and refined estimates of axonal diameter and dispersion. Oscillating gradient spin echo (OGSE) at high frequencies provides access to sub-cellular diffusion time scales for greater insights into tissue microstructure and water exchange. Beyond human imaging, we are now developing ex vivo imaging pipelines integrating diffusion MRI that capitalizes on high b-value diffusion MRI with optical clearing, light-sheet microscopy, and micro-CT to validate tractography and microstructural metrics across scales. By uniting cutting-edge hardware, advanced modeling, and rigorous validation, Connectome 2.0 establishes a unique platform for precision neuroscience, linking cellular architecture to large-scale circuits and illuminating how structure supports function in studies of human behavior, aging, and pathology.

Understanding the brain across scales - from cellular architecture to whole-brain networks - is central to uncovering the principles of brain function and dysfunction. Diffusion MRI offers a unique opportunity to bridge these scales: it is sensitive to microstructural features at the cellular level, yet applicable to non-invasive mapping of connectivity across the entire brain. The challenge lies in interpretation, as inferring complex cellular organisation from MRI signals measured at millimetre resolution is inherently ill-posed. By contrast, microscopy provides rich detail of cellular architecture but is typically restricted to small, ex vivo samples. My research focuses on integrating information from these complementary techniques to overcome their individual limitations and gain new insights into tissue organisation. Rather than relying on simple MRI–microscopy correlations, we are developing "data-fusion" or joint modelling approaches that combine the two modalities directly. Our aim is to enhance diffusion MRI’s ability to bridge scales, enabling more biologically grounded interpretations of macroscale MRI signals. In this talk, I will highlight work from the lab advancing such cross-modal, cross-scale, and cross-species methods for investigating microstructure and connectivity.

Nonhuman primates (NHPs), and in particular macaques, are fundamental for both basic neuroscience and preclinical research due to their close neuroanatomical, physiological, and behavioral similarities to humans. The use of NHPs, when combined with noninvasive imaging techniques, enable direct whole-brain comparisons across species. However, given that the macaque brain is only about 6% the size of the human brain, significantly higher spatial resolution is needed to produce translatable results. Advances in ultrahigh-field MRI here at CMRR have advanced neuroscience research by enabling unprecedented spatial/ temporal resolution and signal sensitivity in NHPs. Our work spans a broad range of topics from cross-modal and cross-species comparative approaches to anatomical connectivity, preclinical longitudinal models of psychiatric disorders, and broader, more abstract neuroscience concepts such as neural and behavioral timescales. A critical component of our research is the implementation of cross-modal comparisons—integrating data from structural MRI, functional MRI, diffusion imaging, viral tract tracing, optical imaging and electrophysiology. These cross-modal analyses allow us to validate and calibrate neural markers across different imaging modalities, bridging the gap between methods and species. Moreover, we have integrated ultrahigh fields in NHPs with innovative methods for dense, longitudinal neurobehavioral sampling. This combination facilitates a comprehensive, multi-modal approach to uncover the neural substrates and progression of neuropsychiatric disorders.  Our goal is to use high-field MRI alongside other methods, leveraging their complementary strengths to enhance research insights, while also striving to better integrate MRI techniques into the animal research field to facilitate more comprehensive studies.