Echo Time Dependence of Laminar BOLD Activation at 7 Tesla
Peter Jan Koopmans¹, Markus Barth^1,2, David Gordon Norris^1,2
1Radboud University Nijmegen, Donders Institute for Brain, Cognition and Behaviour, Nijmegen, Netherlands; ²Erwin L. Hahn Institute for Magnetic Resonance Imaging, Essen, Germany

We present a multi-echo fMRI study at 7 T with 0.75 mm isotropic voxels and TEs ranging from 4.8 to 56 ms. Layer dependent T2* values are reported for human V1 showing a gradient from lower T2* near white matter and higher near the cortical surface with a superimposed dip in the granular layer. We show that the intravascular contribution to GE-BOLD at 7 T is dominated by the pial compartment and that laminar activation profiles are TE dependent. The optimal TE to detect BOLD changes in parenchyma is ~28 ms considerably longer than previously thought as previous estimates have included venous blood.

Retinotopically Organized Left to Right Hemisphere Functional Connectivity in Human V1 Using High-Resolution FMRI at 7T
Jonathan Rizzo Polimeni¹, Kyoko Fujimoto¹, Bruce Fischl^1,2, Douglas N. Greve¹, Lawrence L. Wald^1,3
1Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Harvard Medical School, Massachusetts General Hospital, Charlestown, MA, United States; ²Computer Science and AI Lab (CSAIL), Massachusetts Institute of Technology, Cambridge, MA, United States; ³Harvard-MIT Division of Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, MA, United States

Functional connectivity analysis of resting-state fMRI data has been used to investigate large-scale networks of brain activity. Here investigate whether functional connectivity analysis exhibits sufficient spatial specificity to detect retinotopic organization of the cross-hemispheric correlations detected in cortical area V1. The observed pattern of functional connectivity follows the retinotopic layout—presumably due to the retinotopically-organized common drive from the retina via the LGN. This indicates that despite the indirect nature of these inter-hemispheric connections, an orderly topographic pattern is present and functional connectivity analysis possesses the specificity to detect small-scale organization of the connections within a single cortical area.

Detailed Topographic and Functional Mapping of Areas Within the Posterior Lateral-Occipital and HMT/V5 Complex at 3T Using Functional Grid Analysis
Hauke Kolster¹, Ron Peeters², Guy A. Orban^1
1Lab. for Neuro- and Psychophysiology, KU Leuven, Leuven, Belgium; ²Radiology, UZ Leuven, Leuven, Belgium

We functionally mapped areas within the human posterior lateral-occipital (LOC) and hMT/V5 complex. Using a topographical alignment and correlating retinotopic with unsmoothed functional data we developed a fMRI group analysis, which is specific to within fractions of the individual areas. We demonstrate that the human MT/V5 complex includes the homologue of the macaque MT/V5 field-map cluster, consisting of areas V4t, MT/V5, MSTv, and FST. We further show that these areas can be sharply distinguished from neighboring areas in LOC based on functional characteristics and that a previously reported overlap of motion and shape responses coincides with areas V4t and FST.

Mapping the Early Spatiotemporal BOLD FMRI Response in the Barrel Cortex of Rats
Xin Yu¹, Stephen Dodd¹, Yoshiyuki Hirano¹, Daniel Glen², Ziad S. Saad², Richard C. Reynolds², Afonso C. Silva¹, Alan P. Koretsky^1
1NINDS, NIH, Bethesda, MD, United States; ²NIMH, NIH, Bethesda, MD, United States

BOLD-fMRI signals increase in the rat somatosensory cortex faster than the transit time of blood moving from arteries to veins, which enables us to measure the evolution of BOLD responses at early times after stimulation. Here, the rat barrel cortex activity was mapped at 0.2s temporal resolution in 2D GE-EPI images at 150mmx150μmx500μm using an 11.7T MRI. Activity-evoked BOLD signals were first observed at 0.8s, and shifted to adjacent penetrating venules at 1-1.2s, later propagating to the superficial draining veins. This indicates that BOLD-fMRI maps made prior to about 1 s will have minimal contribution from penetrating cortical venules.

Relative Timing of Brain Activations Revealed by Ultra-Fast MR Inverse Imaging (InI)
Fa-Hsuan Lin¹, Thomas Witzel¹, Tommi Raij, Jyrki Ahveninen, John Bellveau
¹A. A. Martinos Center, Charlestown, MA, United States

We use the ultra fast MR inverse imaging (InI) to interrogate the feasibility to detect hemodynamic timing difference across the brain areas using a two-choice reaction time task. We hypothesize that the vascular response variability can be reduced in the group-level analysis such that neuronally related timing information can become distinct. The MRI and behavior results supported this hypothesize by showing statistically significant timing first at visual and then at motor cortices in our group data (N=23).

Investigation of Seizure Propagation Using EEG-FMRI and Dynamic Causal Modelling
Patricia Figueiredo¹, Alberto Leal^2
1Instituto Superior Técnico, Lisbon, Portugal; ²Department of Neurophysiology, Hospital Júlio de Matos, Lisbon, Portugal

One of the challenges of EEG-fMRI techniques in epilepsy is the investigation of the spatio-temporal dynamics of seizure-related BOLD signals. Here, we have employed Dynamic Causal Modelling (DCM) to test a number of competing models of discharge propagation within a network of functionally connected brain areas identified from EEG-fMRI data of ictal activity, in a patient with epilepsy associated with a hypothalamic hamartoma. Our results demonstrated the feasibility and utility of DCM in the study of the origin and propagation pathway of seizure activity, which may be of critical importance when deciding the surgical approach for epilepsy treatment.

Support Vector Machine Classification of FMRI Data in Image and K-Space Domains
Scott Peltier¹, Jonathan Lisinski², Douglas Noll, Stephen LaConte^2
1Functional MRI Laboratory, University of Michigan, Ann Arbor, MI, United States; ²Computational Psychiatry Unit, Baylor College of Medicine, Houston, TX, United States

This work examines support vector machine (SVM) classification of complex fMRI data, both in the image domain and in the acquired k-space data.  We achieve high classification accuracy using image magnitude, image phase, and k-space magnitude data.  Additionally, we maintain high classification accuracy even when using only partial k-space data.

A Rapid Whole-Brain Classifier for Real-Time Functional MRI Feedback
Jeremy F. Magland¹, Ze Wang², Daniel Willard², Anna Rose Childress^2,3
1Department of Radiology, University of Pennsylvania Medical Center, Philadelphia, PA, United States; ²Department of Psychiatry, University of Pennsylvania Medical Center, Philadelphia, PA, United States; ³VA VISN 4 MIRECC, Philadelphia, PA, United States

Recent studies demonstrate that functional MRI subjects can learn to control activity in localized areas of the brain through the use of real-time fMRI feedback. Potential implications of this technology include a variety of therapies, such as pain management for patients suffering from chronic pain, and craving suppression in individuals with addictions. Whereas much is known about which specific brain regions to target in the case of pain management, less is known about which regions impact craving in addicted individuals. To address this challenge, we have implemented a real-time feedback system based on whole-brain classification.

Hadamard-Encoded FMRI for Reduced Susceptibility Dropout
Gary H. Glover¹, Catherine E. Chang^1
1Radiology, Stanford University, Stanford, CA, United States

The susceptibility difference between air and tissue induces intravoxel dephasing that causes signal dropout in BOLD fMRI. Thin slices can mitigate some of this loss but at a severe SNR efficiency penalty that is only partially offset by summing adjacent slices together. We propose a method that uses Hadamard encoding of two thin subslices per slice subsequently combined incoherently with UNFOLD to recover signal at no loss of SNR in uniform regions. Results using 2 2mm subslices and a hypercapnic challenge demonstrate a 10% increase in activation volume in frontal-orbital regions when compared with conventional 4 mm slice acquisitions.

Rapid Multiecho 3D Radial FMRI
Gregory R. Lee¹, Jean Tkach¹, Mark Griswold^1,2
1Department of Radiology, Case Western Reserve University, Cleveland, OH, United States; ²Biomedical Engineering, Case Western Reserve University, Cleveland, OH, United States

A method to perform multi-echo BOLD functional MRI using an undersampled, multishot 3D radial trajectory is demonstrated.  The proposed view-ordering scheme is a 3D analog of bit-reversed view ordering and allows reconstruction at power of 2 undersampling factors (2,4,8,16).  Aliasing artifacts are periodic in time and can be removed via UNFOLD.  Whole brain images were reconstructed at five echo times (TE=7.3, 16.1, 24.9, 33.6 and 42.4 ms) while maintaining a temporal resolution of 798 ms / volume.  The multiple echoes can used to create dynamic T2* maps and may be combined via weighted summation (optimizing sensitivity over multiple T2* values).