Balu Krishnan¹, Wanyong Shin², Ajay Nemani², Anna Crawford², and Mark Lowe^2
1Epilepsy Center, Cleveland Clinic Foundation, CLEVELAND, OH, United States, ²Mellen Center, Cleveland Clinic Foundation, CLEVELAND, OH, United States

EEG data acquired during simultaneous EEG-fMRI studies are prone to environmental and physiological artifacts. In this study, we detail an EEG artifact reduction pipeline for a block design task paradigm during a BOLD/ASL sequence.  Briefly, the pipeline consists of removal of MR and cardioballistic artifact, ICA based correction for removing eyeblink and residual scanner artifacts, and source modeling of EEG data to remove additional artifacts. Pipeline was tested on 7 subjects (6 multiple sclerosis patients and 1 normal-control). The EEG data processed using the pipeline shows high fidelity and is comparable to similar data acquired outside the scanner.

Wanyong Shin¹, Balu Krishnan², Ajay Nemani¹, Anna Crawford¹, and Mark J Lowe^1
1Radiology, Cleveland Clinic, Cleveland, OH, United States, ²Epilepsy, Cleveland Clinic, Cleveland, OH, United States

Resting state motor network is investigated using EEG theta to beta ratio (TBR) and compared with calibrated fMRI.

Marta Xavier¹, Inês Esteves¹, Athanasios Vourvopoulos¹, Ana R Fouto¹, Amparo Ruiz-Tagle¹, Raquel Gil-Gouveia², and Patrícia Figueiredo^1
1Department of Bioengineering, Institute for Systems and Robotics, Lisbon, Portugal, ²Neurology Department, Hospital da Luz, Lisbon, Portugal

We investigated an integration strategy whereby relevant EEG features are extracted and linear models learnt so as to predict the Default Mode Network activity measured by simultaneous fMRI. We compared the performance of four models: root-mean-squared frequency (RMSF), total power (TP), linear combination of band-specific power (LC), and weighted degree of the functional connectivity network built from the band-specific imaginary part of coherency of the EEG cross-spectrum (WD-ICoh). Models were estimated using elastic net regularization and were found to predict the target BOLD signal with fairly good correlations. Although these varied significantly across models, WD-ICoh outperformed the remaining models.

Rodolfo Abreu¹, Júlia F. Soares¹, Sónia Batista^2,3, Lívia Sousa^2,3, Miguel Castelo-Branco^1,3, and João Valente Duarte^1,3
1Coimbra Institute for Biomedical Imaging and Translational Research (CIBIT), Institute for Nuclear Sciences Applied to Health (ICNAS), University of Coimbra, Coimbra, Portugal, ²Neurology Department, Centro Hospitalar e Universitário de Coimbra, Coimbra, Portugal, ³Faculty of Medicine, University of Coimbra, Coimbra, Portugal

There is no consensus regarding the optimal EEG source reconstruction algorithm, nor a systematic comparison between sets of fMRI-derived spatial priors to be included in the inversion models. We compared four inversion algorithms, each with three sets of fMRI-derived spatial priors consisting of task activation maps, resting-state networks and, for the first time, dynamic functional connectivity states. We found that combining a beamformer with these priors improves the reconstruction quality, especially in terms of the overlap with task-related brain regions of interest and RSN templates. Our results may guide more informatively the selection of the optimal EEG source reconstruction approach.

Inês Esteves¹, Ana R. Fouto¹, Amparo Ruiz-Tagle¹, Athanasios Vourvopoulos¹, Marta Xavier¹, Nuno A. Silva², Raquel Gil-Gouveia³, Agostinho Rosa¹, and Patrícia Figueiredo^1
1ISR-Lisboa and Department of Bioengineering, Instituto Superior Técnico – Universidade de Lisboa, Lisbon, Portugal, ²Learning Health, Hospital da Luz, Lisbon, Portugal, ³Neurology Department, Hospital da Luz, Lisbon, Portugal

Physiological noise correction of the BOLD signal requires information concerning the cardiac and respiratory activity. For EEG-fMRI, the burden of performing extra recordings may be alleviated by retrieving an ECG-derived respiratory signal (EDR). This work aimed to evaluate different methods for EDR estimation and subsequent derivation of fMRI respiratory regressors, by comparison with the measured respiratory signal, for EEG-fMRI. The feasibility of EDR estimation was demonstrated, with KPCA and PCA methods achieving the best performance. Nevertheless, similarities between EDRs and respiration were not directly reflected by the regressors, and the impact of using EDR-based respiratory regressors should be further examined.

Beni Mulyana^1,2, Aki Tsuchiyagaito¹, Jared Smith¹, Masaya Misaki¹, Samuel Cheng², Martin Paulus¹, Hamed Ekhtiari¹, and Jerzy Bodurka^1
1Laureate Institute for Brain Research, Tulsa, OK, United States, ²Electrical and Computer Engineering, University of Oklahoma, Tulsa, OK, United States

We designed a closed-loop concurrent transcranial alternating current stimulation and functional MRI (tACS-fMRI) to optimize fronto-parietal fMRI connectivity. Capacity for modulating fMRI connectivity are highly dependent on the transcranial electric stimulation (tES) parameters such as electric current frequency and phase. We proposed closed-loop tACS-fMRI optimization to search and find the tACS parameters resulting in the highest frontoparietal connectivity. Comparing to control group, experimental group shows an increased task-related frontoparietal functional connectivity (FC) during the training runs. The improvements in working memory was observed in participants in experimental tACS group (testing vs baseline run, 2-back task).

Hsin-Ju Lee^1,2, Mikko Nyrhinen³, Risto J. Ilmoniemi³, and Fa-Hsuan Lin^1,2,3
1Physical Sciences Platform, Sunnybrook Research Institute, Toronto, ON, Canada, ²Department of Medical Biophysics, University of Toronto, Toronto, ON, Canada, ³Department of Neuroscience and Biomedical Engineering, Aalto University, Espoo, Finland

We measured fMRI signals in response to excitatory and inhibitory TMS modulations over the human primary motor cortex (M1) with TMS bursts of high (10- and 30-Hz) and low frequency (0.5-Hz) with a controlled TMS dosage over the 30-s interval, respectively. Excitatory and inhibitory modulations were evidenced by motor evoked potential changes. Significantly increased fMRI signal at M1 was only detected under excitatory high-frequency TMS but not during inhibitory low-frequency TMS. The supplementary motor area (SMA) had significant fMRI signal changes after both kinds of TMS. The topology of the activated M1 and SMA matched those during the voluntary movement.

Thomas Lindner¹, Hajrullah Ahmeti², Michael Helle³, Olav Jansen⁴, Michael Synowitz², and Stephan Ulmer^4,5
1University Hospital Hamburg-Eppendorf, Hamburg, Germany, ²Neurosurgery, University Hospital Schleswig-Holstein, Kiel, Germany, ³Tomographic Imaging Department, Philips Research Laboratories, Hamburg, Germany, ⁴Department of Radiology and Neuroradiology, University Hospital Schleswig-Holstein, Kiel, Germany, ⁵Radiology, Kantonsspital Winterthur, Winterthur, Switzerland

Using EPI Arterial Spin Labeling, resting state networks can be mapped during neurosurgery without an additional acquisition of a BOLD scan.

Thomas Lindner¹, Michael Helle², Olav Jansen³, and Stephan Ulmer^3,4
1University Hospital Hamburg-Eppendorf, Hamburg, Germany, ²Tomographic Imaging Department, Philips Research Laboratories, Hamburg, Germany, ³Department of Radiology and Neuroradiology, University Hospital Schleswig-Holstein, Kiel, Germany, ⁴Radiology, Kantonsspital Winterthur, Winterthur, Switzerland

In this study, the effects of separating the label and control condtion from an Arterial Spin Labeling dataset used for resting state mapping was investigated and no differences between the label and the control condition could be found.

 

Kaden T Shearer¹, Allen A Champagne², Nicole S Coverdale¹, Ingrid S Johnsrude³, and Douglas J Cook^4
1Centre for Neuroscience Studies, Queen's University, Kingston, ON, Canada, ²Department of Medicine, Queen's University, Kingston, ON, Canada, ³The Brain and Mind Institute, University of Western Ontario, London, ON, Canada, ⁴Department of Surgery, Queen's University, Kingston, ON, Canada

Movie fMRI is a promising alternative to resting state paradigms given its ability to evoke predictable patterns of neuronal activation. However, it remains unknown whether increased reliability of cortical activation is accompanied by increased correlation between the BOLD and CBF signals (BOLD-CBF coupling). This study compared the intersubject synchronization of the BOLD signal across 7 functional brain networks between resting state and movie fMRI, as well as potential differences in BOLD-CBF coupling. Movie fMRI significantly improved the reliability of cortical activation between subjects compared to resting state, although no significant differences in BOLD-CBF coupling were documented.

Hyerin Oh^1,2, Ben Babourina-Brooks^1,2,3, Adam Berrington^2,3, Dorothee P Auer^1,2,3, Henryk Faas^1,2, and JeYoung Jung^4
1Division of Clinical Neuroscience, School of Medicine, University of Nottingham, Nottingham, United Kingdom, ²Sir Peter Mansfield Imaging Centre, University of Nottingham, Nottingham, United Kingdom, ³NIHR Nottingham Biomedical Research Centre, Queen’s Medical Centre, University of Nottingham, Nottingham, United Kingdom, ⁴School of Psychology, University of Nottingham, Nottingham, United Kingdom

Dynamic changes in Glx [glutamate + glutamine] and GABA concentrations were investigated in the dorsolateral prefrontal cortex (DLPFC) during a working memory using fMRS and fMRI at 3T. During the 12 minutes 2-back task Glx showed the trend of increase, while GABA levels remained stable. fMRI confirmed the increased regional activity in the DLPFC during 2-back vs. 0-back task. Importantly, the Glx changes were correlated with the DLPFC BOLD regional activity and task performance during 2-back task. Our results suggest that working memory related Glx changes may be involved in the DLPFC’s functional response and performance in working memory.

Tory Frizzell¹, Elisha Phull², Mishaa Khan², Jodie Gawryluk³, Xiaowei Song², and Ryan C.N. D'Arcy^4
1Engineering Science, Simon Fraser University, Surrey, BC, Canada, ²Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, BC, Canada, ³Psychology, University of Victoria, Victoria, BC, Canada, ⁴Computing Science, Simon Fraser University, Surrey, BC, Canada

A gap exists in developing a sensitive method for detecting functional neuroplasticity in white matter. To investigate this, participants trained on a motor learning task for two weeks and were scanned pre and post training using task based BOLD fMRI and DTI. Low frequency oscillations in time series BOLD data demonstrated that average amplitudes decreased with training in the internal capsule and corpus callosum genu. DTI analysis detected white matter neuroplasticity in internal capsule and corona radiata using fractional anisotropy. The distributed effect of motor learning suggests that multi-modal whole brain approaches will provide a more comprehensive understanding neuroplasticity.

Mei-Yu Yeh^1,2, Henry S Chen², Ping Hou², Vinodh A. Kumar³, Jason M Johnson³, Kyle R Noll⁴, Sujit S Prabhu⁵, Donald F. Schomer³, and Ho-Ling Liu ^2
1Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsinchu, Taiwan, ²Department of Imaging Physics, The University of Texas MD Anderson Cancer Center, Houston, Houston, TX, United States, ³Departments of Diagnostic Radiology, The University of Texas MD Anderson Cancer Center, Houston, TX, United States, ⁴Department of Neuro-oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, United States, ⁵5Department of Neurosurgery, The University of Texas MD Anderson Cancer Center, Houston, TX, United States

Clinical functional MRI (fMRI) can be limited by neurovascular uncoupling (NVU) resulted from the pathology. Cerebrovascular reactivity (CVR) mapping based on BOLD MRI during a hypercapnia task, such as breath hold (BH), has been proposed to indicate the potential NUV. However, BH-MRI is limited by patient’s corporation. Recent studies have suggested that CVR measurement can be derived from resting-state (rs) fMRI that has also been increasing utilized to map functional networks in clinical settings. This study evaluates the CVR mapping with rs-fMRI and compare with BH-MRI in patients with gliomas. 

Stefano Moia^1,2, Vicente Ferrer^1,2, Rachael C Stickland³, Ross Davis Markello⁴, Eneko Uruñuela^1,2, Maite Termenon¹, César Caballero-Gaudes¹, and Molly G Bright^3,5
1Basque Center on Cognition, Brain and Language, Donostia, Spain, ²University of the Basque Country UPV/EHU, Donostia, Spain, ³Department of Physical Therapy and Human Movement Sciences, Feinberg School of Medicine, Northwestern University, Chicago, IL, United States, ⁴Montreal Neurological Institute, McGill University, Montreal, QC, Canada, ⁵Biomedical Engineering, McCormick School of Engineering and Applied Sciences, Northwestern University, Chicago, IL, United States

The reliability of cerebrovascular reactivity varies across brain regions, but little is known about what factors drive such variability. By leveraging a high-quality precision functional mapping dataset of breath-hold induced cerebrovascular reactivity (7 subjects with 10 sessions each), we assessed how much parcellations describing different architectures (i.e. vascular anatomy, functional networks, subcortical anatomy) could present internal homogeneity, and hence explain these regionally-specific reliability. We found that all architectures taken into account can equally explain these spatial patterns of cerebrovascular reactivity.

Linqing Li¹, Sean Marrett¹, Andy John Derbyshire¹, and Peter Bandettini ^2
1Functional MRI Facility/NIMH, National Institutes of Health, Bethesda, MD, United States, ²National Institute of Mental Health, National Institutes of Health, Bethesda, MD, United States

Conventional CO₂ calibration for quantification functional changes of CMRO2 may be sensitive to the experimental conditions. We derive a nonlinear coupling relationship of CMRO2 and CBF as simple as (CMRO2/CMRO2₀)=(CBF/CBF₀)^0.25, when α=0.38, β=1.5, where M-value from calibration is not necessary, from Davis model. Nonlinearity between percentage changes of δ(CMRO2)=CMRO2/CMRO2₀-1 and δ(CBF)=CBF/CBF₀-1, in our equation is shown comparable to two classical models. Calculations of δ(CMRO2), based on our equation, are approximately consistent with previous results from both PET and MRI. Additionally, we prove that when desired, M can be measured directly from BOLD and CBF functional changes instead of CO₂ calibration.

Michael Bernier^1,2, Jeorg Peter Pfannmoeller^1,2, Saskia Bollmann³, Avery J.L. Berman^1,2, and Jonathan R Polimeni^1,2,4
1Department of Radiology, A. A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA, United States, ²Radiology, Harvard Medical School, Boston, MA, United States, ³Centre for Advanced Imaging, University of Queensland, Brisbane, Australia, ⁴Division of Health Sciences and Technology, Massachusetts Institute of Technology, Boston, MA, United States

We have developed a “forward-model” method to calculate the extravascular fields surrounding the blood vessels of the brain that accounts for the vessel diameter and orientation and estimates the field change with activation using in vivo measures of vessel anatomy and blood susceptibility.

Avery JL Berman^1,2, Avery JL Berman^1,2, Fuyixue Wang^1,3, Kawin Setsompop^4,5, J. Jean Chen^6,7, and Jonathan R Polimeni^1,2,3
1Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States, ²Department of Radiology, Harvard Medical School, Boston, MA, United States, ³Harvard-MIT Division of Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, MA, United States, ⁴Department of Radiology, Stanford University, Palo Alto, CA, United States, ⁵Department of Electrical Engineering, Stanford University, Palo Alto, CA, United States, ⁶Rotman Research Institute, Baycrest Health Sciences, Toronto, ON, Canada, ⁷Department of Medical Biophysics, University of Toronto, Toronto, ON, Canada

We introduce simulations of the BOLD signal that incorporate realistic image encoding gradients to help better understand the role the acquisition window plays on BOLD contrast. We applied this framework to examine the well-known macrovascular R₂' contamination in spin-echo EPI BOLD signals as a function of readout duration, vessel diameter, and voxel size using randomly oriented cylinders. To understand large-vessel effects, more realistic anatomy is needed; therefore, we added a large pial vein to random microvascular networks to systematically investigate macrovascular contamination along cortical depths and across columns. This framework reproduced experimentally established results related to the acquisition window duration.

Mario Gilberto Baez-Yanez¹, Alex Bhogal¹, Wouter Schellekens¹, Jeroen C.W. Siero^1,2, and Natalia Petridou^1
1Department of Radiology, Center for Image Sciences, University Medical Center Utrecht, Utrecht, Netherlands, ²Spinoza Centre for Neuroimaging Amsterdam, Amsterdam, Netherlands

The administration of CO₂/O₂ gas-challenges during BOLD-fMRI is appealing due to the hemodynamic impact these stimuli have. With the support of computational modeling, arterial gas manipulations can provide a means to separate neuro-vascular signal contributions. Hence, we performed simulations in virtual vascular architectures for different blood volumes and oxygen saturation levels. The effect that these states have on the magnetic field was simulated, separately, for arteries and veins. We present look-up tables to derive the possible vascular contribution responsible for measured BOLD signals in clinical or research fMRI settings including changes in local blood volume, oxygen metabolism or cereberovascular diseases.

Jacob Chausse¹, Avery J. L. Berman², and J. Jean Chen^1,3
1Rotman Research Institute, Baycrest Health Sciences, North York, ON, Canada, ²A. A. Martinos Center for Biomedical Imaging, Harvard Medical School, Massachusetts General Hospital, Boston, MA, United States, ³Department of Medical Biophysics, University of Toronto, Toronto, ON, Canada

Biophysical simulations of the BOLD fMRI phenomenon are indispensable for the development of quantitative fMRI techniques. Since the “gold-standard” 3D Monte Carlo simulation approach was proposed, 2D variations that are less computationally intensive have also been suggested. However, these approaches have never been directly compared. In this work, we directly compare the accuracy of 2D approaches against 3D ones, in order to reconcile results from different simulation studies.