Cesar Caballero Gaudes¹, Ziad S Saad², Mathijs Raemaekers³, Nick F. Ramsey³, and Natalia Petridou^4
1BCBL. Basque Center on Cognition, Brain and Language, Donostia, Guipuzcoa, Spain, ²Statistical and Scientific Computing Core, National Institute of Mental Health, National Institutes of Health, Bethesda, MD, United States, ³Brain Center Rudolf Magnus, Department of Neurology and Neurosurgery, UMC Utrecht, Utrecht, Netherlands, ⁴Radiology, Imaging Division, UMC Utrecht, Utrecht, Netherlands

While most analysis approaches assume temporal stationarity in the study of brain functional connectivity, there is increasing evidence that spontaneous activity in functional networks also comprise more dynamic and transient states. Inferences about dynamic functional connectivity are usually established upon group analyses, thereby inherently excluding the characterization of brain states relating to an individual’s specific cognitive and mental processes. Here, we demonstrate that functional networks can robustly mapped from sparse and brief spontaneous BOLD events in individual subjects by using sparse paradigm free mapping and clustering techniques, as well as benefiting from the high BOLD sensitivity available at 7T and high temporal resolution of 3D-PRESTO

Nishant Zachariah¹, Zhihao Li^2,3, Jason Langley², Shiyang Chen², Mark Davenport¹, Justin Romberg¹, and Xiaoping Hu^2
1Department of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States, ²Department of Biomedical Engineering, Emory University and Georgia Institute of Technology, Atlanta, GA, United States, ³Institute of Affective and Social Neuroscience, Shenzhen University, Shenzhen, Guangdong, China

In this study, we demonstrate a previously undiscovered function of Frontal Pole Cortex(FPC) in the regulation higher order cognitive tasks. We leverage machine learning techniques to data mine state of the art fMRI time series to uncover the role of the FPC. Remarkably, we are able to show that by using the time series of only 4 voxels (of > 900,000), with only a linear classifier, we are able to predict with >90% accuracy which of 7 tasks + resting state activity that a subject was performing. The most common location of these voxels across subjects is in the FPC.

Ruo-Ning Sun¹, Ying-Hua Chu¹, Yi-Cheng Hsu¹, Wen-Jui Kuo², and Fa-Hsuan Lin^1
1Institute of Biomedical Engineering, National Taiwan University, Taipei, Taiwan, ²Institute of Neuroscience, National Yang Ming University, Taipei, Taiwan

The spatial resolution of MR inverse imaging (InI) was empirically tested at 3T and 7T. By using a coil array of the same number of channel and a similar geometry at a higher field, we found that the coil sensitivity becomes more disparate and improves the condition of the spatial encoding. Compared to results at 3T, the InI spatial resolution quantified by the average point-spread function at 7T improved by about 65% and 90% at SNR = 0.1 and 1, respectively.

Jonathan R. Polimeni¹, Marta Bianciardi¹, Boris Keil¹, and Lawrence L. Wald^1,2
1Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Harvard Medical School, Massachusetts General Hospital, Charlestown, Massachusetts, United States, ²Harvard-MIT Division of Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States

Physiological noise fluctuations are driven by several mechanisms, and their effects have been shown to vary across brain regions. Here we investigate the contribution of several physiological noise sources on resting-state BOLD as a function of cortical depth. We find that all physiological measurements, including cardiac, respiratory, and end-tidal CO2, explain a higher percentage of the signal variance in BOLD signal sampled near to the pial surface compared to near the white matter interface. However, a depth-dependent seed-based analysis of the Default Mode Network showed only a modest effect of sampling depth. Target audience: Clinicians/researchers using high resolution fMRI, or studying physiological noise in fMRI signals.

Benedikt A Poser¹ and Desmond HY Tse^1,2
1Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, Netherlands, ²Department of Radiology, Maastricht University, Maastricht, Netherlands

Slice-specific spokes pulses were designed for and applied to high resolution 2D EPI imaging at 9.4T in order to mitigate the signal and SNR variations across the brain due to RF inhomogeneity as typically encountered at ultra-high field. Improvements in signal homogeneity are demonstrated when using the slice-specific three-spokes RF pulses instead of the coil’s CP mode excitations. The B1+ shimmed EPI sequence is applied to BOLD fMRI scans.

Swati Rane¹, Brandon Ally², Emily Mason², Subechhya Pradhan³, Erin Hussey², Kevin Waddell³, Hanzhang Lu^4,5, and Manus Donahue^2,3
1Radiology and Radiological Sciences, Vanderbilt University Institute of Imaging Science, Nashville, TN, United States, ²Neurology, Vanderbilt University, Nashville, TN, United States, ³Radiology and Radiological Sciences, Vanderbilt University, Nashville, TN, United States, ⁴Radiology, UT Southwestern, Dallas, TX, United States, ⁵Psychiatry, UT Southwestern, Dallas, TX, United States

We investigated the relation between brain neurotransmitter concentrations, venous oxygen saturation, and oxygen extraction fraction. We show that venous oxygen saturation is inversely proportional to the ratio of GABA/Glx.

Shiyang Chen¹, Jason Langley¹, and Xiaoping Hu^1
1The Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, Atlanta, GA, United States

Most dynamic functional connectivity analyses are performed using sliding window correlation. One problem is that a fixed sliding window with a predefined length selected ad hoc is used even though the temporal duration of the states is now known to vary. In order to address this challenge, we introduced a Gaussian Hidden Markov Model to model brain state transition with the time series of the fMRI data (in contrast to the method which models the functional connectivity states). This model allows us to detect the spatial patterns of states and the transition sequences of the states. In our study, we detected 9 reproducible brain states as combination of conventional resting state networks.

David Provencher¹, Andreas Bartels², Yves Bérubé-Lauzière^3,4, and Kevin Whittingstall^4,5
1Department of Nuclear Medicine and Radiobiology, Université de Sherbrooke, Sherbrooke, QC, Canada, ²Werner Reichardt Centre for Integrative Neuroscience, Tübingen, Germany, ³Department of Electrical and Computer Engineering, Université de Sherbrooke, Sherbrooke, QC, Canada, ⁴Centre d'imagerie moléculaire de Sherbrooke (CIMS), Université de Sherbrooke, Sherbrooke, QC, Canada, ⁵Department of Diagnostic Radiology, Université de Sherbrooke, Sherbrooke, QC, Canada

Typical task-fMRI studies aim to compute brain activation maps through voxel-wise correlation of measured and modeled BOLD timecourses. This usually relies on two hypotheses, namely that 1) neural activity follows the stimulation waveform (e.g. a boxcar function) and that 2) the hemodynamic response function (HRF), relating neural and BOLD activity, follows a canonical model. Here, we acquired sequential EEG-fMRI data in 5 subjects viewing multiple repetitions of a sinusoidally modulated visual stimulus over 8 seconds. Through data analysis and linear deconvolution of HRFs, we show that both hypotheses are inappropriate here, and are therefore not generalizable to slowly changing stimuli.

Zachary B Rodgers¹, Erin K Englund², Maria A Fernandez-Seara³, and Felix W Wehrli^1
1Radiology, University of Pennsylvania, Philadelphia, PA, United States, ²Department of Bioengineering, University of Pennsylvania, Philadelphia, PA, United States, ³Neuroimaging Laboratory, Center for Applied Medical Research, University of Navarra, Pamplona, Navarra, Spain

We present a new approach for calibrated BOLD fMRI in which BOLD and CBF are measured alongside direct quantification of global venous oxygen saturation (Y_v) via susceptometry-based oximetry. The proposed method allows for determination of M, the BOLD calibration factor, without assumptions regarding the CMRO₂ or flow changes associated with the stimulus. Three subjects completed a hypercapnic gas mixture breathing paradigm, with M-values generated from the proposed Y_v-based method in agreement with the traditional Davis model. In one subject, similar M-values were generated from hyperoxia, without the usual requirement of end-tidal O₂ monitoring or the need to assume baseline Y_v.

Daniel Bulte¹, Hannah Hare¹, Nazneen Sudhan², Joanna Simpson², Joseph Donnelly², Xiuyun Liu², and Jonathan Coles^2
1FMRIB, University of Oxford, Oxford, Oxfordshire, United Kingdom, ²WBIC, University of Cambridge, Cambridge, Cambridgeshire, United Kingdom

A dual-gas calibrated MRI paradigm designed to measure multiple cerebrovascular parameters was directly compared in 15 healthy subjects to triple oxygen PET and NIRS cerebral oximetry. CBF measures from pCASL and 15-O PET were found to be well correlated. A weak correlation was found between MRI OEF and 15-O PET OEF, but no correlation was found with NIRS OEF and either of the other modalities.

Hsiao-Ying Wey¹, Jacob M Hooker¹, Michael S. Placzek^1,2, Bruce R Rosen¹, and Joseph B Mandeville^1
1A. A. Martinos Center, Department of Radiology, Massachusetts General Hospital, Harvard Medical School, Charlestown, MA, United States, ²McLean Hospital, Harvard Medical School, Belmont, MA, United States

In this study, we present simultaneous PET/MRI study with pharmacological (μ-opioid receptor antagonist and agonist) challenges in nonhuman primates to determine the relationship between opioid receptor occupancy, dopamine modulation, and changes in CBV. PET and CBV signals show dose-dependent reductions to opioid antagonist challenges. PET and fMRI demonstrated concurrent and overlapping changes in the basal forebrain; however, the largest occupancy changes (PET) were observed in the thalamus and caudate, while the largest CBV changes were observed in the putamen. Taken together with the μ-opioid agonist challenge results, our data suggest that opioid-evoked a dopaminergic component in the measured fMRI signal.

Laurentius Huber¹, Dimo Ivanov², Maria Guidi¹, Robert Turner¹, Kâmil Uludağ², Harald E Möller¹, and Benedikt A Poser^2
1Max Planck Institute for Human Cognitive & Brain Sciences, Leipzig, Germany, ²Maastricht Brain Imaging Centre, Netherlands

In this study, we combined high-field (7 T) VASO with a simultaneous-multi-slice (SMS) acquisition scheme. We implemented and evaluated this method for simultaneous acquisition of functional changes in cerebral blood volume and BOLD signal at high spatial resolution, using a visuo-motor task, and covering a large number of brain areas including V1, V5, M1, and S1. We show that SMS-VASO can measure blood volume changes with high spatial resolution and satisfactory CNR, across multiple brain regions and with better local specificity than GE-BOLD signal.

Haiguang Wen¹ and Zhongming Liu^1,2
1Electrical and Computer Engineering, Purdue University, West Lafayette, Indiana, United States, ²Biomedical Engineering, Purdue University, West Lafayette, Indiana, United States

To elucidate the neural basis of resting state fMRI, we separated and characterized the fractal and oscillatory components of neurophysiological signals observed with electrocorticography (ECoG) and magnetoencephalography (MEG), and evaluated the distinct contributions of such electrophysiological components to resting state fMRI networks by using simultaneously acquired fMRI and electroencephalography (EEG). We found that the globally synchronized fMRI signals were correlated with the fractal component of electrophysiology, and that the fMRI activities of spatially specific networks were coupled to the oscillatory components of electrophysiology. The global fMRI and fractal electrophysiology likely result from common neural modulation pathways with diffusive projects that innervate the entire cortex.

Tung-Lin Wu¹, Feng Wang^1,2, Li Min Chen^1,2, Adam W. Anderson^1,2, Zhaohua Ding^1,2, and John C. Gore^1,2
1Vanderbilt University Institute of Imaging Science, Nashville, TN, United States, ²Radiology and Radiological Sciences, Vanderbilt Univeristy, Nashville, TN, United States

Recently, we reported that anisotropic correlations between resting state signals within a local region of white matter can be used to drive functional structures that closely resemble DTI data but without the use of diffusion gradients. Indeed, we demonstrated a technique that delineates the functional architecture of the brain, especially white matter, purely on the basis of fMRI data. In order to explore and verify the biophysical mechanisms for the observed spatio-temporal correlations in white matter signals, we carried out imaging studies on live anesthetized squirrel monkeys and compared spatio-temporal correlation tensors from T2* and cerebral blood volume (CBV)-weighted fMRI.

Hongfu Sun¹, Peter Seres¹, and Alan H. Wilman^1
1Biomedical Engineering, University of Alberta, Edmonton, Alberta, Canada

We investigate the conditions under which subcortical GM structural QSM can be extracted from standard fMRI experiments enabling brain iron studies at no time cost. We examine the effects of spatial resolution and time series variation in both structural and functional QSM in relation to standard BOLD magnitude fMRI at 1.5 and 4.7 T, and propose a structural QSM reconstruction pipeline for use in standard fMRI studies.