The glymphatic system contributes to waste clearance in the brain and helps in maintaining neural health. This study aimed to evaluate the age-related changes in magnetic resonance imaging (MRI) measurements of the index of diffusivity along the perivascular space (Along the Perivascular Space (ALPS) index) in infants and children less than 5 years of age. This IRB-approved, HIPAA-compliant study utilized diffusion-weighted MRI data from the Baby Connectome Project (BCP), which scanned developing subjects between birth and 5 years. We randomly selected subjects older than 2 months without significant imaging artifacts. The ALPS index was computed using diffusion tensor imaging (DTI)-derived diffusivity metrics along projection and association fibers. Multivariable linear regression and ANOVA were used to assess the impact of age, sex, and motion on ALPS index values across hemispheres and age groups. We analyzed 60 cases (58

Despite numerous magnetic resonance imaging (MRI) head motion mitigation strategies, the lack of rigorous evaluation limits their optimization and clinical adoption. We propose an in vivo framework combining a visual instruction system for reproducible head motion with reference standard interpose displacement estimation to assess intra-MRI tracking accuracy and precision. Its utility is demonstrated by comparing a markerless optical system (MOS) and a fat-signal navigator (FatNav). Six participants underwent 3T T1-weighted brain MRI with a FatNav module, performing visually guided 2° and 4° head rotations around the X- and Z-axes using MOS feedback. T1-weighted images were acquired at seven distinct head poses. MOS and FatNav motion estimates were compared against rigid registration of the T1-weighted images, which served as the reference standard. MOS- and FatNav-corrected images for the three successive head rotations were also compared using the structural similarity index measure (SSIM), peak signal-to-noise ratio (PSNR), and a focus measure. FatNav accuracy was inferior for translations (p < 0.001) and 2°-4° rotations but improved to match MOS for subtle pitch+ and yaw+, even surpassing it for subtle yaw-. Meanwhile, MOS precision was higher for yaw+ than yaw- (p < 0.001) but inferior to FatNav for pitch+ (p = 0.041). MOS better restored T1-weighted image fidelity, yielding higher SSIM, PSNR, and focus (p < 0.01). Notably, the framework detected a subtle improvement in FatNav performance with neck masking, an effect uncaptured by conventional image quality metrics. In conclusion, while image quality metrics suggested superior overall correction with MOS, our framework provided a more detailed characterization of in vivo performance differences.

Motivation: Measuring neurochemical dynamics, particularly the concentrations of GABA and glutamate, at the single-subject level remains challenging due to the low SNR when a small number of transients are used in the averaging window. Goal(s): To assess the impact of different averaging window sizes on the stability of GABA and glutamate concentration estimates during the time-course of a visual stimulation experiment. Approach: We analyzed the data using various averaging window sizes to evaluate their effect on concentration stability. Results: Detecting single-subject changes in Glutamate and GABA remains challenging, even with a window size of 16 representing the largest practical window size in our experiments. Impact: In single-subject fMRS, choosing averaging window sizes with sufficient SNR is important for reliable measurements of glutamate and GABA dynamics. Further technical development both on the analysis side (temporal fitting) and acquisition side (real-time frequency drift correction) should be explored.

Purpose:Head-motion tracking and correction remains a key area of research in MRI, but the lack of rigorous and standardized evaluation approaches hinders their optimization and comparison. We introduce an in-vivo framework for assessing the accuracy of intra-MRI head motion tracking, and demonstrates its effectiveness by comparing two methods based on a markerless optical system (MOS) and a fat signal navigator (FatNav). Methods:Six participants underwent 3T brain MRI using a T1-weighted (T1w) pulse-sequence with a fat- navigator module. Participants performed head-rotations of 2° or 4°, each visually guided by MOS feedback around a single primary axis (X or Z). MOS and FatNav estimations were evaluated against rigid-registration of T1w-images, as gold-standard, across seven different head positions. Results:The proposed approach revealed that MOS outperforms FatNav in estimating translation and large head rotations (2-4°), while FatNav shows better accuracy for subtle rotations. Image quality assessments following correction for three head rotations (rightward, upward, and leftward) confirmed that MOS outperformed FatNav in restoring image fidelity, as evidenced by the higher Structural Similarity Index, Peak Signal-to-Noise Ratio, and Focus Measure. Unlike the traditional image quality- based comparisons, the proposed framework demonstrated sensitivity to subtle improvements in FatNav performance, achieved by applying a neck mask to the fat-navigator images. Conclusion:The proposed framework enabled a precise in-vivo evaluation and comparison of MOS and FatNav for head-motions estimation. It was sufficiently sensitive to reveal a slight improvement in FatNav performance when neck was masked in fat-navigator images. In parallel, the conventional image quality-based approach confirmed the superior performance of MOS in restoring T1W image quality, though it did not capture the improvement achieved by FatNav with neck-masking. Together, these two complementary approaches provide a comprehensive assessment of both head-motions estimation and correction in MRI.

Motivation: Classical 1D linescan acquisitions have recently been shown to be valuable for recording in-vivo MRI signals across the layers of human cerebral cortex; however, these reduced-field-of-view techniques are vulnerable to rotational head movements as well as translations parallel and perpendicular to the line. Goal(s): To acquire high-quality, high-resolution linescan data that is robust to "in-line" and "through-line" motion. Approach: 3D-EPI volumetric navigators (vNavs) were incorporated into a spin-echo-based linescan pulse sequence. Results: We demonstrate that, by combining prospective and retrospective motion correction, we can acquire reliable linescan data with 0.5-mm readout resolution at 7T, in the presence of in-line and through-line head motion. Impact: Motion-robust linescan techniques will help enable the measurement of tissue microstructure and microvascular fMRI signals at high spatial resolutions, approaching the thickness of individual cortical layers, facilitating noninvasive studies of cortical circuitry and architectonics in the living human brain.

∆B0 shim optimization performed at the beginning of an MR scan is unable to correct for ∆B0 field inhomogeneities caused by patient motion or hardware instability during scans. Navigator-based methods have been demonstrated previously to be effective for motion and shim correction. The purpose of this work was to accelerate volumetric navigators to allow fast acquisition of the parent navigated sequence with short real-time feedback time and high spatial resolution of the ∆B0 field mapping. A GRAPPA-accelerated 3D dual-echo EPI vNav was implemented on a 3 T Prisma MRI scanner. Testing was performed on an anthropomorphic head phantom and 11 human participants. vNav-derived ∆B0 field maps with various spatial resolutions were compared to Cartesian-encoded gold-standard 3D gradient-echo ∆B0 field mapping. ∆B0 shimming was evaluated for the scanner's spherical harmonics shims and a custom-made AC/DC RF-receive/∆B0-shim array. The performance of dual-echo and single-echo accelerated navigators was compared for tracking and updating ∆B0 field maps during motion. Real-time motion and shim corrections for 2D MRI and 3D MRSI sequences were assessed in vivo with controlled head movement. Up to 8-fold acceleration of volumetric navigators (vNavs) significantly reduced geometric distortions and signal dropouts near air-tissue interfaces and metal implants. Acceleration allowed a flexible tradeoff between spatial resolution (2.5-7.5 mm) and acquisition time (242-1302 ms). Notably, accelerated high-resolution (5 mm) vNav was faster (378 ms) than unaccelerated low-resolution (7.5 mm) vNav (700 ms) and showed better agreement with 3D-GRE ∆B0 field mapping with 5.5 Hz RMSE, 1 Hz bias, and [-10%, +10%] confidence interval. Accelerated vNavs improved 3D MRSI and 2D MRI in real-time motion and shim correction applications. Advanced shimming with spherical harmonic and shim array showed superior ΔB0 correction, especially with joint shim optimization. GRAPPA-accelerated vNavs provide fast, robust, and high-quality ∆B0 field mapping and shimming over the whole-brain. The accelerated vNavs enable rapid correction of ∆B0 field inhomogeneities and faster acquisition of the navigated parent sequence. This methodology can be used for real-time motion and shim correction to enhance data quality in various MRI applications.

Motivation: High-resolution time-of-flight (TOF) magnetic resonance angiography (MRA) is needed for imaging the cerebral vasculature at the mesoscopic scale but is challenged by involuntary head movements during image acquisition. Goal(s): This study tests whether whole-brain volumetric navigators (vNavs), based on 3D-EPI, can be used for prospective motion correction (PMC) of thin-slab TOF acquisitions at 7T. Approach: Interactions between vNavs and TOF sequence modules are examined, and the performance of PMC is investigated under different head motion conditions. Results: While results show that vNavs affect TOF signal, vNav motion correction enables the robust imaging of small vessels. Impact: We show that slab-selective isotropic 0.16-mm TOF-MRA is feasible with whole-brain volumetric navigators for prospective head motion correction, enabling robust in vivo imaging of the human vasculature at an unprecedented scale.

Diffusion MRI of the infant brain allows investigation of the organizational structure of maturing fibers during brain development. Post-mortem imaging has the potential to achieve high resolution by using long scan times, enabling precise assessment of small structures. Technical development for post-mortem diffusion MRI has primarily focused on scanning of fixed tissue, which is robust to effects like temperature drift that can cause unfixed tissue to degrade. The ability to scan unfixed tissue in the intact body would enable post-mortem studies without organ donation, but poses new technical challenges. This paper describes our approach to scan setup, protocol optimization, and tissue protection in the context of the Developing Human Connectome Project (dHCP) of neonates. A major consideration was the need to preserve the integrity of unfixed tissue during scanning in light of energy deposition at ultra-high magnetic field strength. We present results from one of the first two subjects recruited to the study, who died on postnatal day 46 at 29+6 weeks postmenstrual age, demonstrating high-quality diffusion MRI data. We find altered diffusion properties consistent with post-mortem changes reported previously. Preliminary voxel-wise and tractography analyses are presented with comparison to age-matched in vivo dHCP data. These results show that high-quality, high-resolution post-mortem data of unfixed tissue can be acquired to explore the developing human brain.

Accurate labeling of specific layers in the human cerebral cortex is crucial for advancing our understanding of neurodevelopmental and neurodegenerative disorders. Building on recent advancements in ultra-high-resolution ex vivo MRI, we present a novel semi-supervised segmentation model capable of identifying supragranular and infragranular layers in ex vivo MRI with unprecedented precision. On a dataset consisting of 17 whole-hemisphere ex vivo scans at 120 $\mu $m, we propose a Multi-resolution U-Nets framework that integrates global and local structural information, achieving reliable segmentation maps of the entire hemisphere, with Dice scores over 0.8 for supra- and infragranular layers. This enables surface modeling, atlas construction, anomaly detection in disease states, and cross-modality validation while also paving the way for finer layer segmentation. Our approach offers a powerful tool for comprehensive neuroanatomical investigations and holds promise for advancing our mechanistic understanding of progression of neurodegenerative diseases.

Motivation:To develop a sequence for high-quality concurrent measurement of BOLD signal changes (fMRI) and biochemicals metabolite concentrations (fMRS). Goals:To implement parallel imaging with inline image reconstruction to improve fMRI image quality in concurrent fMRI-fMRS experiments at 7 T. Approach:We modified an fMRS-fMRI sequence to start by acquiring reference lines for GRAPPA reconstruction. Then each TR consists of a semiLASER acquisition for single-voxel MRS, and a GRAPPA-accelerated 3D EPI acquisition. Results:We obtained sufficient tSNR (30) map for the 3D EPI and a high SNR (59) and a narrow linewidth (9 Hz) for the spectrum. Impact:The modified concurrent fMRI-fMRS pulse sequence enhances the fMRI component to enable whole-brain coverage, reduced distortion, and high spatial resolution, providing a powerful tool for neuroscientists to study the dynamics of neurochemicals simultaneously with the BOLD signal.

Motivation: Many cutting-edge MR neuroimaging paradigms require real-time decision making and precise FOV positioning. We present two software tools to support such paradigms. Goal(s): Develop two modules. 1) vSend: opens a socket and sends imaging data to another computer in a vendor-agnostic format, enabling real-time analysis. 2) AAhijack: reads a matrix from a socket and overwrites the Siemens AutoAlign matrix, enabling online slice prescription. Approach: Modules are implemented as Siemens image reconstruction modules (ICE functors) in C++ and two slice prescription systems utilizing the modules are demonstrated. Results: The slice prescription systems have comparable performance and various advantages and disadvantages. Impact: The software tools presented have enabled a variety of cutting-edge MR neuroimaging paradigms including real-time fMRI, motion tracker calibration, real-time shimming, fetal head-pose detection and automated FOV prescription, reacquisition planning and single-slice BOLD imaging FOV prescription.

Motivation: Very high quality of MR spectroscopic imaging (MRSI) data is needed for robust and reproducible metabolite quantification. This critically depends on the B0 shimming and scan stability. Integrated RF-receive/B0-shim arrays significantly improve spectral quality. Goal(s): Real-time motion correction and multicoil shimming update with an integrated RF-receive/B0-shim array for robust whole-brain MRSI. Approach: We developed a rapid navigator for head tracking and B0 fieldmapping in combination with rapid processing for real-time update of multicoil shim currents and MRSI localization. Results: Real-time motion correction and multicoil shimming provides significantly narrower linewidth, higher signal-to-noise, reduced quantification errors and reproducible metabolic imaging. Impact: Whole-brain MRSI is a unique method for non-invasive mapping of brain neurochemistry, and in combination with real-time motion correction and multicoil shim array update provides robust and reproducible quantitative metabolic imaging for clinical use.

Compressedsensing magnetic resonance imaging (CS-MRI) seeks to recover visual information from subsampled measurements for diagnostic tasks. Traditional CS-MRI methods often separately address measurement subsampling, image reconstruction, and task prediction, resulting in a suboptimal end-to-end performance. In this work, we propose Tackle as a unified co-design framework for jointly optimizing subsampling, reconstruction, and prediction strategies for the performance on downstream tasks. The naïve approach of simply appending a task prediction module and training with a task-specific loss leads to suboptimal downstream performance. Instead, we develop a training procedure where a backbone architecture is first trained for a generic pre-training task (image reconstruction in our case), and then fine-tuned for different downstream tasks with a prediction head. Experimental results on multiple public MRI datasets show that Tackle achieves an improved performance on various tasks over traditional CS-MRI methods. We also demonstrate that Tackle is robust to distribution shifts by showing that it generalizes to a new dataset we experimentally collected using different acquisition setups from the training data. Without additional fine-tuning, Tackle leads to both numerical and visual improvements compared to existing baselines. We have further implemented a learned 4×-accelerated sequence on a Siemens 3 T MRI Skyra scanner. Compared to the fully-sampling scan that takes 335 seconds, our optimized sequence only takes 84 seconds, achieving a four-fold time reduction as desired, while maintaining high performance.

We propose a wave-encoded model-based deep learning (wave-MoDL) method for joint multi-contrast image reconstruction with volumetric encoding using an interleaved look-locker acquisition sequence with T 2 preparation pulse (3D-QALAS). Wave-MoDL enables a 2-minute acquisition at R=4x3-fold acceleration using a 32-channel array to provide T 1 , T 2 , and proton density maps at 1 mm isotropic resolution, from which standard contrast-weighted images can also be synthesized.

Intra-scan motion costs tens of thousands of dollars per scanner annually due to the need to repeat non-diagnostic scans1. When assessing the scale of the problem and potential solutions, radiologists’ ratings of artifacts are considered the gold standard. However, inconsistent and conflicting ratings must be consolidated into a single gold-standard. We introduce a hierarchical label fusion algorithm that infers each rater's performance and promotes consistency across slices from a volume. This algorithm reduces label noise compared to majority votes, and allows non-expert ratings to be calibrated and included as additional silver-standards.

We demonstrate a deep learning approach for fast retrospective intraslice rigid motion correction in segmented multislice MRI. A hypernetwork uses auxiliary rigid motion parameter estimates to produce a reconstruction network based on the motion parameters that are specific to the input image. This strategy produces higher quality reconstructions than those produced by model-based techniques or by networks that do not use motion estimates. Further, this approach mitigates sensitivity to misestimation of the motion parameters.

Brain cells are arranged in laminar, nuclear, or columnar structures, spanning a range of scales. Here, we construct a reliable cell census in the frontal lobe of human cerebral cortex at micrometer resolution in a magnetic resonance imaging (MRI)–referenced system using innovative imaging and analysis methodologies. MRI establishes a macroscopic reference coordinate system of laminar and cytoarchitectural boundaries. Cell counting is obtained with a digital stereological approach on the 3D reconstruction at cellular resolution from a custom-made inverted confocal light-sheet fluorescence microscope (LSFM). Mesoscale optical coherence tomography enables the registration of the distorted histological cell typing obtained with LSFM to the MRI-based atlas coordinate system. The outcome is an integrated high-resolution cellular census of Broca’s area in a human postmortem specimen, within a whole-brain reference space atlas.

Motion artifacts can negatively impact diagnosis, patient experience, and radiology workflow especially when a patient recall is required. Detecting motion artifacts while the patient is still in the scanner could potentially improve workflow and reduce costs by enabling immediate corrective action. We demonstrate in a clinical k-space dataset that using cross-correlation between adjacent phase-encoding lines can detect motion artifacts directly from raw k-space in multi-shot multi-slice scans. We train a split-attention residual network to examine the performance in predicting motion artifact severity. The network is trained on simulated data and tested on real clinical data.

Subject motion can cause artifacts in clinical MRI, frequently necessitating repeat scans. We propose to alleviate this inefficiency by predicting artifact scores from partial multi-shot multi-slice acquisitions, which may guide the operator in aborting corrupted scans early.