Biology subjects

Setsompop, K.

Publications and source records attributed to Setsompop, K..

15 recordsLinked to original sources

bioRxiv2026

Joint reconstruction of multi-contrast relaxation-diffusion imaging

Diffusion MRI across multiple gradient directions, b-values and echo times (TEs) provides complementary tissue contrast but requires extensive sampling. Joint k-space and q-space undersampling requires reconstruction that preserves diffusion- and TE-dependent signal variation while exploiting shared anatomical information. Here we introduce DIFFUSE (Diffusion-relaxometry Implicit Field Fitting from Undersampled k-q Space across Echoes), a subject-specific joint k-q-TE image reconstruction framework. It fits a single implicit neural representation (INR) directly to complementary undersampled multi-coil k-space measurements across diffusion encodings and echo times. The INR (a self-supervised neural network) maps spatial location, b-value and TE to baseline b0 signals and spherical harmonic coefficients for directional diffusion-weighted image reconstruction. The measurement-specific forward model incorporates phase, coil sensitivities and motion and eddy current related transforms. We evaluated DIFFUSE using retrospective Human Connectome Project data, controlled motion simulations and in vivo multi-shell and multi-TE acquisitions. At sixfold in-plane acceleration, DIFFUSE reduced normalized mean squared error by 57-73% relative to SENSE across two diffusion shells in the retrospective experiment. In the controlled motion experiment, motion-aware DIFFUSE reduced normalized mean squared error by 66-72% relative to static DIFFUSE. The prospective study demonstrated multi-shell reconstruction at 1.3 mm isotropic resolution with significantly high signal-to-noise ratio (SNR). In a separate multi-TE experiment, a single representation jointly reconstructed all nine contrasts corresponding to 3 b-value/TE combinations. DIFFUSE thus provides a common reconstruction strategy for accelerated protocols spanning multiple diffusion encodings and echo times.

neuroscience↗

bioRxiv2026

High-SNR Whole-Brain Mesoscale Diffusion MRI Using Rotating-View Acquisitions (ROVER-dMRI)

Purpose: We aim to acquire and reconstruct high signal-to-noise ratio (SNR), whole-brain, mesoscale (isotropic submillimeter) diffusion MRI (dMRI) in vivo in a clinically feasible scan time. Methods: We present ROVER-dMRI, a unified rotating-view acquisition and super-resolution reconstruction framework. Thick slices (up to 10 mm)acquired across many rotated orientations improve the base SNR, and a shared multi-resolution hash-encoded implicit neural representation (INR) with per-direction MLP head and a physics-informed, Bloch-simulated slice-profile forward model reconstructs these views into a continuous isotropic volume (up to 13x super-resolution) in a self-supervised manner. ROVER-dMRI was evaluated in vivo on human brain (0.78 mm isotropic and 0.5 mm isotropic) and monkey brain (0.6 mm isotropic) across MRI platforms with varying gradient performance, and benchmarked against gSlider-SMS and least-squares super-resolution reconstruction (LS-SRR). Results: At 0.78 mm isotropic resolution, ROVER-dMRI diffusion- weighted images achieved higher SNR than the vendor product sequence, whose images were limited by the noise floor, while its directionally encoded color (DEC) maps showed lower noise than gSlider. When reducing the number of views from 20 to 13 (reducing the scan time by 65% ), ROVER-dMRI acquisition/reconstruction preserved all fine details in the fiber orientations. In the monkey brain, its DEC map matched a 60-direction gSlider acquisition at only 40% of the scan time. ROVER-dMRI preserved coherent fiber-orientation-distribution (FOD) structure, including crossing fibers, and recovered fine details in the cerebellum at 0.5 mm isotropic, that are not visible at 0.78 mm resolution. Compared to state-of- the-art 0.5 mm methods, ROVER-dMRI reduced scan time by 73% while acquiring more diffusion directions. Conclusion: ROVER-dMRI offers a practical route to high-SNR whole-brain mesoscale diffusion imaging in vivo, with the pulse sequence and reconstruction pipeline released as open source. KEYWORDS: diffusion MRI, super-resolution, implicit neural representation, hash encoding, physics informed learning

neuroscience↗

bioRxiv2026

Postnatal maturation of putamen microstructure accompanies topographic white matter connectivity and altered circuits in autism

The putamen is a major hub of the basal ganglia that emerges early in gestation. However, whether its mature organization is established before birth or emerges postnatally remains unknown. Using cross-sectional and longitudinal quantitative MRI (R1 and R2*, related to tissue density and iron, respectively) and diffusion MRI, we characterized the development of putamen's microstructure and its white matter connectivity with cortex from birth to 12 months and compared their trajectories with those in adults. Despite its prenatal emergence, the putamen undergoes substantial postnatal development. R1 increases from birth to 12 months, producing a prominent anterior-posterior gradient, whereas R2* increases primarily between age one and adulthood, producing a medial-lateral gradient. Cortico-putamen white matter connectivity is diffuse in infants but becomes topographic in adults, with anterior putamen linked to frontal cortex and posterior putamen to sensorimotor cortex. In autism spectrum disorder, this organization is largely preserved and accompanied by increased anterior putamen-prefrontal connectivity. Our findings reveal distinct spatial developmental trajectories of putamen microstructure and cortical connectivity providing a developmental framework for understanding the organization of the putamen in infancy, which has implications for assessing neurodevelopmental disorders of the basal ganglia.

neuroscience↗

bioRxiv2026

Field-Correcting GRAPPA (FCG): a technique to correctspatiotemporal-varying phase errors in Echo Planar Imaging

PurposeTo develop a Field-Correcting GRAPPA (FCG) technique to correct the spatiotemporal-varying phase errors in EPI caused by eddy currents. MethodsThe fast-changing gradient in EPI causes strong eddy current effects and associated spatiotemporal-varying phase errors, producing significant image artifacts. The use of higher gradient amplitude, slew rate, and ramp sampling factor for faster imaging exacerbates this problem. In this work, FCG was developed to address this challenge by using a multi-layer perceptron (MLP) to provide a compact representation of a family of GRAPPA-like kernels that correct the spatiotemporal-varying phase errors in the data. A dedicated calibration pipeline was designed to acquire high-quality source and target data for MLP training in both slice-by-slice and simultaneous multi-slice (SMS) acquisitions. To validate FCGs assumptions and performance, a field camera was used to provide ground-truth measurement of phase patterns. The performance of FCG was further validated on phantom and in vivo experiments using demanding EPI trajectories across multiple 3T and 7T systems. ResultsField camera measurements revealed strong spatiotemporal phase variations along the kx direction that repeat along ky during EPI readouts. The experiments on high-performance systems across 3T and 7T demonstrate that FCG can provide superior correction for the artifacts induced by spatiotemporal-varying phase errors compared with existing approaches. ConclusionFCG is an effective and robust method for correcting spatiotemporal phase errors in EPI, enabling improved image quality on high-performance systems.

bioengineering↗

bioRxiv2026

Scout-based Multi-Echo NAvigator (SMENA) for high temporal resolution motion and B0 estimation and correction: applications to multi-echo GRE and EPTI

PurposeTo develop a data-driven technique, Scout-based Multi-Echo NAvigator (SMENA), for joint estimation of motion and B0 inhomogeneity ({delta}B0) at a temporal resolution of [~]200 ms with minimal additional scan time for gradient-echo acquisition. MethodsSMENA consists of two key acquisition components: SMENA-scout and SMENA-nav. SMENA-scout is a rapid 3D 4-mm multi-echo acquisition completed in less than 8 seconds, providing images with matched contrast and phase at multiple echo times. SMENA-nav captures signal variations induced by motion and{delta} B0 during the scan using compact multi-echo navigator trajectories (3.5 ms) embedded within each TR. Motion and{delta} B0 maps were jointly estimated every [~]200 ms through a model-based optimization framework relating SMENA-scout to SMENA-nav. The estimation accuracy and correction performance of SMENA were evaluated in simulations and in vivo using multi-echo GRE and GRE-EPTI acquisitions. Multiple prospective motion experiments, including large continuous movement and deep breathing, were investigated. ResultsIn both simulations and in vivo experiments, accurate motion and{delta} B0 estimation were achieved. Compared with motion-only estimation, joint estimation reduced rotation and translation errors. Joint motion and{delta} B0 correction resulted in substantial improvements in image quality, particularly at longer echo times, producing an NRMSE of 10.4% compared to 31.6% with motion-only correction. High-temporal-resolution tracking of motion and{delta} B0 enabled improved reconstruction quality in scenarios involving continuous motion and deep breathing. ConclusionSMENA enables high-temporal-resolution joint estimation of motion and{delta} B0 with minimal additional acquisition cost, providing a practical solution for motion- and{delta} B0-robust MRI.

neuroscience↗

bioRxiv2026

Multiphasic myelination and dendritic growth modulate qMRI signals in human visual cortex.

How does myelin develop in human visual cortex? By combining immunohistochemistry with in vivo and postmortem magnetic resonance imaging of longitudinal relaxation rate (R1), which increases with myelin content, we find that myelin and R1 increase across development but follow distinct trajectories. Immunohistochemistry reveals two phases of myelination: an infant phase of limited oligodendrogenesis, with myelin restricted to deep cortical layers, followed by widespread myelination across all layers during childhood. Cortical R1 also increases across development and correlates with myelin by childhood. However, in infancy, R1 increases outpace myelin growth and instead tracks dendritic arborization, indicating that the microstructural drivers of R1 change across development. We hypothesize that deep layer myelination in infancy contributes to early visual function whereas later myelination of superficial layers enables prolonged cortical plasticity and learning of complex visual behaviors.

neuroscience↗

bioRxiv2024

Fast and reliable quantitative measures of white matter development with magnetic resonance fingerprinting

Developmental cognitive neuroscience aims to shed light on evolving relationships between brain structure and cognitive development. To this end, quantitative methods that reliably measure individual differences in brain tissue properties are fundamental. Standard qualitative MRI sequences are influenced by scan parameters and hardware-related biases, and also lack physical units, making the analysis of individual differences problematic. In contrast, quantitative MRI can measure physical properties of the tissue but with the cost of long scan durations and sensitivity to motion. This poses a critical limitation for studying young children. Here, we examine the reliability and validity of an efficient quantitative multiparameter mapping method - Magnetic Resonance Fingerprinting (MRF) - in children scanned longitudinally. We focus on T1 values in white matter, since quantitative T1 values are known to primarily reflect myelin content, a key factor in brain development. Forty-nine children aged 8-13y (mean 10.3y {+/-}1.4) completed two scanning sessions 2-4 months apart. In each session, two 2-minute 3D-MRF scans at 1mm isotropic resolution were collected to evaluate the effect of scan duration on image quality and scan-rescan reliability. A separate calibration scan was used to measure B0 inhomogeneity and correct for bias. We examined the impact of scan time and B0 inhomogeneity correction on scan-rescan reliability of values in white matter, by comparing single 2-min and combined two 2-min scans, with and without B0-correction. Whole-brain voxel-based reliability analysis showed that combining two 2-min MRF scans improved reliability (pearsons r=0.87) compared with a single 2-min scan (r=0.84), while B0-correction had no effect on reliability in white matter (r=0.86 and 0.83 4-min vs 2-min). Using diffusion tractography, we delineated MRF-derived T1 profiles along major white matter fiber tracts and found similar or higher reliability for T1 from MRF compared to diffusion parameters (based on a 10-minute dMRI scan). Lastly, we found that T1 values in multiple white matter tracts were significantly correlated with age. In sum, MRF-derived T1 values were highly reliable in a longitudinal sample of children and replicated known age effects. Reliability in white matter was improved by longer scan duration but was not affected by B0-correction, making it a quick and straightforward scan to collect. We propose that MRF provides a promising avenue for acquiring quantitative brain metrics in children and patient populations where scan time and motion are of particular concern.

neuroscience↗

bioRxiv2024

Uncovering microstructural architecture from histology

Mapping the brains fiber network is crucial for understanding its function and malfunction, but resolving nerve trajectories over large fields of view is challenging. Here, we show that computational scattered light imaging (ComSLI) can map fiber networks in histology independent of sample preparation, also in formalin-fixed paraffin-embedded (FFPE) tissues including whole human brain sections. We showcase this method in new and archived, animal and human brain sections, for different sample preparations (in paraffin, deparaffinized, various stains, unstained fresh-frozen). We convert microscopic orientations to microstructure-informed fiber orientation distributions (FODs). Adapting tractography tools from diffusion magnetic resonance imaging (dMRI), we trace axonal trajectories revealing white and gray matter connectivity. These allow us to identify altered microstructure or deficient tracts in demyelinating or neurodegenerating pathology, and to show key advantages over dMRI, polarization microscopy, and structure tensor analysis. Finally, we map fibers in non-brain tissues, including muscle, bone, and blood vessels, unveiling the tissues function. Our cost-effective, versatile approach enables micron-resolution studies of intricate fiber networks across tissues, species, diseases, and sample preparations, offering new dimensions to neuroscientific and biomedical research.

pathology↗

bioRxiv2024

Spherical Echo-Planar Time-resolved Imaging (sEPTI) for rapid 3D quantitative T2* and Susceptibility imaging

PurposeTo develop a 3D spherical EPTI (sEPTI) acquisition and a comprehensive reconstruction pipeline for rapid high-quality whole-brain submillimeter T2* and QSM quantification. MethodsFor the sEPTI acquisition, spherical k-space coverage is utilized with variable echo-spacing and maximum kx ramp-sampling to improve efficiency and incoherency when compared to existing EPTI approaches. For reconstruction, an iterative rank-shrinking B0 estimation and odd-even high-order phase correction algorithms were incorporated into the reconstruction to better mitigate artifacts from field imperfections. A physics-informed unrolled network was utilized to boost the SNR, where 1-mm and 0.75-mm isotropic whole-brain imaging were performed in 45 and 90 seconds, respectively. These protocols were validated through simulations, phantom, and in vivo experiments. Ten healthy subjects were recruited to provide sufficient data for the unrolled network. The entire pipeline was validated on additional 5 healthy subjects where different EPTI sampling approaches were compared. Two additional pediatric patients with epilepsy were recruited to demonstrate the generalizability of the unrolled reconstruction. ResultssEPTI achieved 1.4 x faster imaging with improved image quality and quantitative map precision compared to existing EPTI approaches. The B0 update and the phase correction provide improved reconstruction performance with lower artifacts. The unrolled network boosted the SNR, achieving high-quality T2* and QSM quantification with single average data. High-quality reconstruction was also obtained in the pediatric patient using this network. ConclusionsEPTI achieved whole-brain distortion-free multi-echo imaging and T2* and QSM quantification at 0.75 mm in 90 seconds which has the potential to be useful for wide clinical applications.

bioengineering↗

bioRxiv2023

Deep Learning Initialized Compressed Sensing (Deli-CS) in Volumetric Spatio-Temporal Subspace Reconstruction

IntroductionSpatio-temporal MRI methods enable whole-brain multi-parametric mapping at ultra-fast acquisition times through efficient k-space encoding, but can have very long reconstruction times, which limit their integration into clinical practice. Deep learning (DL) is a promising approach to accelerate reconstruction, but can be computationally intensive to train and deploy due to the large dimensionality of spatio-temporal MRI. DL methods also need large training data sets and can produce results that dont match the acquired data if data consistency is not enforced. The aim of this project is to reduce reconstruction time using DL whilst simultaneously limiting the risk of deep learning induced hallucinations, all with modest hardware requirements. MethodsDeep Learning Initialized Compressed Sensing (Deli-CS) is proposed to reduce the reconstruction time of iterative reconstructions by "kick-starting" the iterative reconstruction with a DL generated starting point. The proposed framework is applied to volumetric multi-axis spiral projection MRF that achieves whole-brain T1 and T2 mapping at 1-mm isotropic resolution for a 2-minute acquisition. First, the traditional reconstruction is optimized from over two hours to less than 40 minutes while using more than 90% less RAM and only 4.7 GB GPU memory, by using a memory-efficient GPU implementation. The Deli-CS framework is then implemented and evaluated against the above reconstruction. ResultsDeli-CS achieves comparable reconstruction quality with 50% fewer iterations bringing the full reconstruction time to 20 minutes. ConclusionDeli-CS reduces the reconstruction time of subspace reconstruction of volumetric spatio-temporal acquisitions by providing a warm start to the iterative reconstruction algorithm.

neuroscience↗

bioRxiv2021

SNR-efficient distortion-free diffusion relaxometry imaging using ACcelerated Echo-train shifted EPTI (ACE-EPTI)

PurposeTo develop an efficient acquisition technique for distortion-free diffusion MRI and diffusion-relaxometry. MethodsA new ACcelerated Echo-train shifted Echo-Planar Time-resolved Imaging (ACE-EPTI) technique is developed to achieve high-SNR, distortion- and blurring-free diffusion and diffusion-relaxometry imaging. ACE-EPTI employs a newly designed variable density spatiotemporal encoding with self-navigation capability, that allows submillimeter in-plane resolution using only 3-shot. Moreover, an echo-train-shifted acquisition is developed to achieve minimal TE, together with an SNR-optimal readout length, leading to ~30% improvement in SNR efficiency over single-shot EPI. To recover the highly accelerated data with high image quality, a tailored subspace image reconstruction framework is developed, that corrects for odd/even-echo phase difference, shot-to-shot phase variation, and the B0 field changes due to field drift and eddy currents across different dynamics. After the phase-corrected subspace reconstruction, artifacts-free high-SNR diffusion images at multiple TEs are obtained with varying T2* weighting. ResultsSimulation, phantom and in-vivo experiments were performed, which validated the 3-shot spatiotemporal encoding provides accurate reconstruction at submillimeter resolution. The use of echo-train shifting and optimized readout length improves the SNR-efficiency by 27-36% over single-shot EPI. The reconstructed multi-TE diffusion images were demonstrated to be free from distortion (susceptibility and eddy currents) and phase/field variation induced artifacts. These improvements of ACE-EPTI enable improved diffusion tensor imaging and rich multi-TE information for diffusion-relaxometry analysis. ConclusionACE-EPTI was demonstrated to be an efficient and powerful technique for high-resolution diffusion imaging and diffusion-relaxometry, which provides high SNR, distortion- and blurring-free, and time-resolved multi-echo images by a fast 3-shot acquisition.

bioengineering↗

bioRxiv2021

Simultaneous pure T2 and varying T2'-weighted BOLD fMRI using Echo Planar Time-resolved Imaging (EPTI) for mapping laminar fMRI responses

Spin-echo (SE) BOLD fMRI has high microvascular specificity, and thus provides a more reliable means to localize neural activity compared to conventional gradient-echo BOLD fMRI. However, the most common SE BOLD acquisition method, SE-EPI, is known to suffer from T2' contrast contamination with undesirable draining vein bias. To address this, in this study, we extended a recently developed distortion/blurring-free multi-shot EPI technique, Echo-Planar Time-resolved Imaging (EPTI), to cortical-depth dependent SE-fMRI at 7T to test whether it could provide purer SE BOLD contrast with minimal T2' contamination for improved neuronal specificity. From the same acquisition, the time-resolved feature of EPTI also provides a series of asymmetric SE (ASE) images with varying T2' weightings, and enables extraction of data equivalent to conventional SE EPI with different echo train lengths (ETLs). This allows us to systematically examine how T2'-contribution affects different SE acquisition strategies using a single dataset. A low-rank spatiotemporal subspace reconstruction was implemented for the SE-EPTI acquisition, which incorporates corrections for both shot-to-shot phase variations and dynamic B0 drifts. SE-EPTI was used in a visual task fMRI experiment to demonstrate that i) the pure SE image provided by EPTI results in the highest microvascular specificity; ii) the ASE EPTI series, with a graded introduction of T2' weightings at time points farther away from the pure SE, show a gradual sensitivity increase along with increasing draining vein bias; iii) the longer ETL seen in conventional SE EPI acquisitions will induce more draining vein bias. Consistent results were observed across multiple subjects, demonstrating the robustness of the proposed technique for SE-BOLD fMRI with high specificity.

neuroscience↗

bioRxiv2021

Ultrafast high-resolution multi-parametric brain MRI using 3D Echo Planar Time-resolved Imaging

Multi-parametric quantitative MRI has shown great potential to improve the sensitivity and specificity of clinical diagnosis and to enhance our understanding of complex brain processes, but suffers from long scan time especially at high spatial resolution. To address this long-standing challenge, we introduce a novel approach, termed 3D Echo Planar Time-resolved Imaging (3D-EPTI), which significantly increases the acceleration capacity of MRI sampling, and provides high acquisition efficiency for multi-parametric MRI. This is achieved by exploiting the spatiotemporal correlation of MRI data at multiple timescales through new encoding strategies within and between efficient continuous readouts. Specifically, an optimized spatiotemporal CAIPI encoding within the readouts combined with a radial-block sampling strategy across the readouts enables an acceleration rate of 800 folds in the k-t space. A subspace reconstruction was employed to resolve thousands of artifact-free high-quality multi-contrast images spaced at a time interval of ~1 ms. We have demonstrated the ability of 3D-EPTI to provide robust and repeatable whole-brain simultaneous T1, T2, T2*, PD and B1+ mapping at high isotropic resolution within minutes (e.g., 1-mm isotropic resolution in 3 minutes), and to enable submillimeter multi-parametric imaging to study detailed brain structures. HighlightsO_LIUltra-fast acquisition for 3D multi-parametric quantitative MRI. C_LIO_LISimultaneous T1 T2 T2* PD and B1+ mapping. C_LIO_LI3-minute scan at 1-mm isotropic resolution with whole-brain coverage. C_LIO_LIMulti-parametric mapping at 700-m isotropic resolution in 10 minutes. C_LIO_LIRepeatable quantification and cortical-depth analysis. C_LI

bioengineering↗

bioRxiv2021

High-fidelity fast volumetric brain MRI using a hybrid denoising generative adversarial network (HDnGAN)

PurposeReducing scan times is important for wider adoption of high-resolution volumetric MRI in research and clinical practice. Emerging fast imaging and deep learning techniques provide promising strategies to accelerate volumetric MRI without compromising image quality. In this study, we aim to leverage an advanced fast imaging technique, wave-controlled aliasing in parallel imaging (Wave-CAIPI), and a novel denoising generative adversarial network (GAN) to achieve accelerated high-fidelity, high-signal-to-noise-ratio (SNR) volumetric MRI. Methods3D T2-weighted fluid-attenuated inversion recovery (FLAIR) image data were acquired on 33 multiple sclerosis (MS) patients using a prototype Wave-CAIPI sequence (acceleration factor R=3x2, 2.75 minutes) and a standard T2-SPACE FLAIR sequence (R=2, 7.25 minutes). A hybrid denoising GAN entitled "HDnGAN" composed of a 3D generator (i.e., a modified 3D U-Net entitled MU-Net) and a 2D discriminator was proposed to denoise Wave-CAIPI images with the standard FLAIR images as the target. HDnGAN was trained and validated on data from 25 MS patients by minimizing a combined content loss (i.e., mean squared error (MSE)) and adversarial loss with adjustable weight{lambda} , and evaluated on data from 8 patients unseen during training. The quality of HDnGAN-denoised images was compared to those from other denoising methods including AONLM, BM4D, MU-Net, and 3D GAN in terms of their similarity to standard FLAIR images, quantified using MSE and VGG perceptual loss. The images from different methods were assessed by two neuroradiologists using a five-point score regarding sharpness, SNR, lesion conspicuity, and overall quality. Finally, the performance of these denoising methods was compared at higher noise levels using simulated data with added Rician noise. ResultsHDnGAN effectively denoised noisy Wave-CAIPI images with sharpness and rich textural details, which could be adjusted by controlling{lambda} . Quantitatively, HDnGAN ({lambda}=10-3) achieved low MSE (7.43 x10-4{+/-}0.94x10-4) and the lowest VGG perceptual loss (1.09x10-2{+/-}0.18x10-2). The reader study showed that HDnGAN ({lambda}=10-3) significantly improved the SNR of Wave-CAIPI images (4.19{+/-}0.39 vs. 2.94{+/-}0.24, P<0.001), outperformed AONLM (4.25{+/-}0.56 vs. 3.75{+/-}0.90, P=0.015), BM4D (3.31{+/-}0.46, P<0.001), MU-Net (3.13{+/-}0.99, P<0.001) and 3D GAN ({lambda}=10-3) (3.31{+/-}0.46, P<0.001) regarding image sharpness, and outperformed MU-Net (4.21{+/-}0.67 vs. 3.29{+/-}1.28, P<0.001) and 3D GAN ({lambda}=10-3) (3.5{+/-}0.82, P=0.001) regarding lesion conspicuity. The overall quality score of HDnGAN ({lambda}=10-3) (4.25{+/-}0.43) was significantly higher than those from Wave-CAIPI (3.69{+/-}0.46, P=0.003), BM4D (3.50{+/-}0.71, P=0.001), MU-Net (3.25{+/-}0.75, P<0.001), and 3D GAN ({lambda}=10-3) (3.50{+/-}0.50, P<0.001), with no significant difference compared to standard FLAIR images (4.38{+/-}0.48, P=0.333). The advantages of HDnGAN over other methods were more obvious at higher noise levels. ConclusionHDnGAN provides robust and feasible denoising while preserving rich textural detail in empirical volumetric MRI data and is superior on both quantitative and qualitative evaluation compared to the original Wave-CAIPI images and images denoised using standard methods. HDnGAN concurrently benefits from the improved image synthesis performance of the 3D convolution and the increased number of samples for training the 2D discriminator from a limited number of subjects. Our study supports the use of HDnGAN in combination with modern fast imaging techniques such as Wave-CAIPI to achieve high-fidelity fast volumetric MRI.

neuroscience↗

bioRxiv2020

In vivo human whole-brain Connectom diffusion MRI dataset at 760 μm isotropic resolution

We present a whole-brain in vivo diffusion MRI (dMRI) dataset acquired at 760 m isotropic resolution and sampled at 1260 q-space points across 9 two-hour sessions on a single healthy subject. The creation of this benchmark dataset is possible through the synergistic use of advanced acquisition hardware and software including the high-gradient-strength Connectom scanner, a custom-built 64-channel phased-array coil, a personalized motion-robust head stabilizer, a recently developed SNR-efficient dMRI acquisition method, and parallel imaging reconstruction with advanced ghost reduction algorithm. With its unprecedented resolution, SNR and image quality, we envision that this dataset will have a broad range of investigational, educational, and clinical applications that will advance the understanding of human brain structures and connectivity. This comprehensive dataset can also be used as a test bed for new modeling, sub-sampling strategies, denoising and processing algorithms, potentially providing a common testing platform for further development of in vivo high resolution dMRI techniques. Whole brain anatomical T1-weighted and T2-weighted images at submillimeter scale along with field maps are also made available.

neuroscience↗