View Presentation Video

Thomas R Barrick¹, Carson Ingo^2,3, and Franklyn A Howe^1
1Department of Neurosciences, St George's, University of London, London, United Kingdom, ²Department of Neurology, Northwestern University, Chicago, IL, United States, ³Department of Physical Therapy, Human Movement Sciences, Northwestern University, Chicago, IL, United States

Keywords: Signal Modeling, Diffusion/other diffusion imaging techniques

We show the quasi-diffusion model describes the dMRI signal from low b-value (stretched exponential) to high b-value (power law) regimes via a single, parsimonious function of two independent parameters. We identify new tissue contrast via a signal inflection point indicating the transition to the localisation regime. Quasi-Diffusion Imaging (QDI) parameter estimates converge to stable values as maximum b-value is increased, suggesting quasi-diffusion is a valid model for brain tissue dMRI signal decay. Accuracy of QDI parameters computed from 4 b-values of a 12 b-value acquisition indicate QDI measures may be derived from data acquired in clinically feasible times.

View Presentation Video

Keywords: Signal Modeling, Diffusion/other diffusion imaging techniques

High performance gradients allow for the exploration of an expanded diffusion parameter space, that simplifies biophysical model at ultra-high b=7-30 ms/μm². The choice of b-encoding space can suppress contributions from extra-axonal water signal while utilization of high performance gradient systems allows for maintaining short echo times (TE<63 ms) with adequate SNR. In this study, the feasibility and reproducibility of mapping effective axonal diameter distributions in the in-vivo brain was assessed by making use of a test-retest paradigm. Whole brain white-matter and parcel based reproducibility and sensitivity were evaluated for this promising biomarker.

View Presentation Video

Misha Pieter Thijs Kaandorp^1,2, Peter Thomas While^1,2, and Frank Zijlstra^1,2
1Department of Radiology and Nuclear Medicine, St. Olav's University Hospital, Trondheim, Norway, ²Department of Circulation and Medical Imaging, NTNU – Norwegian University of Science and Technology, Trondheim, Norway

Keywords: Signal Modeling, Diffusion/other diffusion imaging techniques

DWI data is spatially homogeneous, yet microstructures are irregular, where neighboring voxels do not always share information. Therefore, we need to utilize neighboring correlations only when they are present. In simulations, we show that by training on synthetic data with all plausible combinations of neighboring correlations, the accuracy of supervised deep learning IVIM model fitting can be improved. Conversely, unsupervised learning did not benefit from incorporating spatial information. In in-vivo data from a glioma patient, supervised training on this synthetic data improved the performance of IVIM fitting by effectively denoising the DWI data while preserving edge-like structures.

View Presentation Video

Lipeng Ning^1,2, Carl-Fredrik Westin^1,2, and Yogesh Rathi^1,2
1Brigham and Women's Hospital, Boston, MA, United States, ²Harvard Medical School, Boston, MA, United States

Keywords: Signal Representations, Diffusion/other diffusion imaging techniques

We proposed a method for joint modeling and analysis for diffusion MRI data with multiple TEs and examined the performance of using in vivo data acquired from a clinical scanner. The contribution of this work includes the characterization of b-value-dependent T₂ relaxation rates, comparing prediction results using subsampled TEs, and characterization of microstructural spectral and long and short TEs using extrapolated signals. The results show that the proposed method can provide novel features to characterize T₂-dependent microstructure using rdMRI acquired from clinical scanners.

View Presentation Video

Evren Özarslan¹, Deneb Boito¹, and Alfredo Ordinola^1
1Department of Biomedical Engineering, Linköping University, Linköping, Sweden

Keywords: Signal Modeling, Diffusion/other diffusion imaging techniques

In 1966, Kac asked the question ‘Can one hear the shape of a drum?’ which refers to linking the density of states function to the shape of the pores. We illustrate how the density of eigenstates as well as the pore shape can be obtained through diffusion measurements performed via a recently introduced arrangement of long and narrow gradient pulses. In the presence of dispersity, maps derived from the signal provide exquisite details about the ensemble of pore shapes within the voxel.

View Presentation Video

Fabian Küppers^1,2, Seong Dae Yun¹, Philipp Lohmann¹, Christian Filss^1,3, Gabriele Stoffels¹, Karl-Josef Langen^1,3, and Nadim Jon Shah^1,4,5,6
1Institute of Neuroscience and Medicine - 4, Forschungszentrum Juelich GmbH, Jülich, Germany, ²RWTH Aachen University, Aachen, Germany, ³Department of Nuclear Medicine, RWTH Aachen University Hospital, Aachen, Germany, ⁴JARA-BRAIN - Translational Medicine, Aachen, Germany, ⁵Institute of Neuroscience and Medicine - 11, Forschungszentrum Juelich GmbH, Jülich, Germany, ⁶Department of Neurology, RWTH Aachen University Hospital, Aachen, Germany

Keywords: Multi-Contrast, Oxygenation, Oxygen Extraction Fraction

The versatile information provided by simultaneous multi-contrast GE/SE acquisitions keeps interest in this area high. This work extends the previously published 10-echo GE-SE EPIK to investigate brain tumor patients. Within this scope, a single-slice 12-second acquisition provides quantification of T₂ and T₂* with application to oxygen extraction fraction (OEF) information. Data from four tumor patients were acquired, revealing increased T₂/T₂* values and increased OEF in regions with variable amino acid uptake in O-(2-[18F]fluoroethyl)-L-tyrosine (FET) PET.

View Presentation Video

Keywords: Signal Modeling, Multi-Contrast, Gradient-echo spin-echo (GESE); Echo Planar;

The bias and precision of T2 and T2* estimates for a five-echo GESE model were studied using CRLB and Monte Carlo simulations. Different SNR cases were implemented by varying the standard deviation of added noise or the T2/T2* ratio. An in-vivo study with high SNRs for all EPI readouts was performed to validate the simulation.

 

View Presentation Video

Ting Gong¹, Chantal MW Tax^2,3, Matteo Mancini⁴, Derek K Jones⁴, Hui Zhang¹, and Marco Palombo^4,5
1Centre for Medical Image Computing & Department of Computer Science, University College London, London, United Kingdom, ²Cardiff University Brain Research Imaging Centre, School of Physics and Astronomy, Cardiff University, Cardiff, United Kingdom, ³University Medical Center Utrecht, Utrecht University, Utrecht, Netherlands, ⁴Cardiff University Brain Research Imaging Centre, School of Psychology, Cardiff University, Cardiff, United Kingdom, ⁵School of Computer Science and Informatics, Cardiff University, Cardiff, United Kingdom

Keywords: Signal Modeling, Diffusion/other diffusion imaging techniques

This study models the echo time (TE) dependence of the apparent soma and neurite signal fractions derived from the soma and neurite density imaging (SANDI) and quantifies the apparent compartmental T2 relaxation times in the grey matter (GM). The SANDI model divides the intra-cellular space into intra-neurite and intra-soma compartments to account for contribution of water diffusion in cell bodies. By collecting multi-TE SANDI datasets, we provide the first ever estimates of apparent intra-neurite and intra-soma T2 relaxation times in the human brain and characterise their distribution in cortical and subcortical GM regions.

View Presentation Video

Anne Slawig^1,2, Juliana Bibiano², and Herbert Köstler^2
1University Clinic and Outpatient Clinic for Radiology, University Hospital Halle (Saale), Halle (Saale), Germany, ²Department of Diagnostic and Interventional Radiology, University Hospital Würzburg, Würzburg, Germany

Keywords: Signal Modeling, Relaxometry, T1 mapping, inversion recovery, model based reconstruction

This study proposes a combined fit of an inversion-prepared and non-prepared measurement for robust fast T1-mapping. Therefore, 24 slices in the brain were acquired with and without a global inversion pulse before each slice and no waiting time was heeded in between. Results of exponential model reconstructions for inversion measurement only, non-prepared measurement only and combined fitting of both measurements were compared towards their performance for T1-mapping. Robust and accurate T1-maps were achieved by the proposed combined model even in cases of imperfect inversion. It can thus eliminate long waiting times in between inversions or the necessity of perfect inversion pulses.

View Presentation Video

Rémi Dagenais¹ and Georgios D. Mitsis^2
1Biological and Biomedical Engineering, McGill University, Montréal, QC, Canada, ²Bioengineering, McGill University, Montréal, QC, Canada

Keywords: Data Acquisition, fMRI, Physiological denoising, ABP regulation

The relationship between arterial blood pressure (ABP) fluctuations and the BOLD-fMRI signal is mostly unknown due to the limited availability of reliable MR-compatible ABP devices. Here, we propose a combined experimental and mathematical modeling methodology that uses two calibration recordings outside the scanner to estimate the ABP during the MR scan. Significant correlations between BOLD and ABP were observed in all tested subjects, which also revealed that the global BOLD signal is intrinsically correlated with the underlying ABP fluctuations. Our method is expected to improve our understanding of ABP regulation in the brain and allow for improved physiological fMRI denoising.

View Presentation Video

Thomas Gladytz¹, Kathleen Cantow², Jason Michael Millward^1,3, Sonia Waiczies^1,3, Erdmann Seeliger², and Thoralf Niendorf^1,3
1Berlin Ultrahigh Field Facility (B.U.F.F.), Max Delbrück Center for Molecular Medicine in the Helmholtz Association, Berlin, Germany, ²Institute of Translational Physiology, Charité - Universitätsmedizin Berlin, Berlin, Germany, ³Experimental and Clinical Research Center, a joint cooperation between the Charité Medical Faculty and the Max Delbrück Center for Molecular Medicine in the Helmholtz Association, Berlin, Germany

Keywords: Oxygenation, Quantitative Imaging

Quantitative MRI of renal hypoxia, an early key feature in acute kidney injury, is a valuable tool for diagnostics and pathophysiological studies. En route to MR oximetry we propose a biophysical model that uses the kidney size as an additional MRI derivable information to correct for blood volume confounders of oxygenation sensitive T₂^* maps. The model is demonstrated in a preclinical study involving three different vascular occlusions highlighting the confounding effect of blood volume changes and its correction.

View Presentation Video

Jiahao Li^1,2, Pablo Villar-Calle³, Jinwei Zhang^1,2, Chao Li^2,4, Thanh D. Nguyen², Yi Wang^1,2, Jiwon Kim³, Jonathan W. Weinsaft³, and Pascal Spincemaille^2
1Biomedical Engineering, Cornell University, Ithaca, NY, United States, ²Radiology, Weill Cornell Medicine, New York, NY, United States, ³Medicine, Weill Cornell Medicine, New York, NY, United States, ⁴Applied and Engineering Physics, Cornell University, Ithaca, NY, United States

Keywords: Oxygenation, Motion Correction

A free-breathing 3D stack-of-spiral data acquisition was developed for motion-free cardiac quantitative susceptibility mapping, to tackle the challenge where even a short breath-hold is difficult for patients to perform. ECG and respiratory bellow signal were recorded for retrospective motion binning. A 5D dataset incorporating additional cardiac and respiratory phase dimensions were reconstructed with joint spatiotemporal regularization to generate a motion-free cardiac QSM. In healthy volunteers, this method was compared with a motion-averaged reconstruction and with a separate breath-hold spiral cardiac QSM using Cartesian navigator QSM as reference. Equivalent right-to-left heart chamber differential blood oxygenation was observed among all method studied.

View Presentation Video

Rushdi Zahid Rusho¹, Brad H. Story², David Meyer³, Mathews Jacob⁴, and Sajan Goud Lingala^1,5
1Roy J. Carver Department of Biomedical Engineering, The University of Iowa, Iowa City, IA, United States, ²Speech, Language and Hearing Sciences, University of Arizona, Tucson, AZ, United States, ³Janette Ogg Voice Research Center, Shenandoah University, Winchester, VA, United States, ⁴Electrical and Computer Engineering, The University of Iowa, Iowa City, IA, United States, ⁵Department of Radiology, The University of Iowa, Iowa City, IA, United States

Keywords: Signal Modeling, Head & Neck/ENT

The vocal tract encompasses the airspace from the glottis (the space between the vocal folds) to the external lips. This irregular tube filters the glottal sound source, and is modulated by many structures (e.g. the tongue, lips, and velum) to produce speech sounds (e.g. vowels and consonants). In this work, we determine the preliminary feasibility of integrating vocal tract area functions derived from a recently proposed accelerated pseudo-3D dynamic speech MRI scheme  to a parametric  tube talker model to synthesize speech.

View Presentation Video

Keywords: Signal Modeling, Spectroscopy

Performing simultaneous MRS from two-regions is often desirable in spectroscopy studies. However, achieving an optimal B0 shim in two regions is difficult, leading to poor data quality. We investigate a model-based reconstruction for two-voxel MRS, incorporating B0 information, to reconstruct data with reduced T2* decay. We test this approach using least squares and regularization approaches for Hadamard encoding in simulations and phantom. Despite a noise penalty for B0-correction (2-4)-fold, FIDs from two voxels better matched un-broadened spectra under conditions of sufficient SNR. Regularization also produced time-domain reconstructions with less noise than a least-squares approach.

View Presentation Video

Zifei Liang¹, Choong Heon Lee¹, Jennifer A. Minteer², Yongsoo Kim², and Jiangyang Zhang^1
1Radiology, NYU Langone health, new york, NY, United States, ²Penn State University, Hershey, PA, United States

Keywords: Signal Representations, Brain, MR-histology

To infer cellular-level information from MR signals with high sensitivity and specificity is a challenging task. We previously demonstrated the feasibility of mapping myelin in the mouse brain based on multi-contrast MRI using deep learning, but the results were based on limited histological data and MRI data from separate cohorts. In this study, we acquired serial 2-photon and MRI data from the same mice and trained a neural network for mapping myelin. Our results demonstrated enhanced sensitivity and specificity compared to conventional MRI myelin markers, our previous network, and polynomial fitting.