fMRI: Connectivity & Networks

13:30         3595.     Correlation Modulation Networks to Assess Changes in Functional Connectivity with Task Difficulty

Computer 51

Mayuresh S. Korgaonkar¹, Susan Fiore¹, Candice J. Perkins¹, Nancy K. Squires¹, Zengmin Yan¹, Mark E. Wagshul¹

¹Stony Brook University, Stony Brook, New York, USA

Correlation modulation network analyses allow the evaluation of condition specific functional interactions between brain regions. Our previous results show that there is differential recruitment of brain regions in a task difficulty modulated verb generation task. Using the correlation modulation approach, we sought to evaluate the following: 1. If the connectivity between language regions is modulated by experimental conditions – particularly between generating verbs as compared to passive viewing of non-words and difficult as compared to easy to generate verbs; and 2. How these condition-specific modulations of connectivity differ between young and old subjects.

14:00         3596.     Effect of Hemodynamic Spatial Variability on Granger-Based Long Term Causality

Computer 51

Gopikrishna Deshpande¹, George Andrew James¹, Xiaoping Hu¹

¹Georgia Institute of Technology and Emory University, Atlanta, Georgia, USA

Causal relationships between cortical regions can be inferred from fMRI using Granger causality (GC). However, GC may be affected by spatial variability of the hemodynamic response (HRF). We examine the effect of HRF variability on the causal influences in a neural network that evolves slower than the HRF variability and TR. Simulations showed that the HRF variability leads to erroneous results. Furthermore, we show that the true causality could be recovered by using summary measures derived from the BOLD data instead of the raw fMRI time series. These findings are illustrated using data obtained from a fatigue motor task.

14:30         3597.     Network Reduction for Interpreting Large Scale Brain Networks

Computer 51

Gopikrishna Deshpande¹, Randall Stilla, Krish Sathian, Xiaoping Hu¹

¹Georgia Institute of Technology and Emory University, Atlanta, Georgia, USA

In this work, we illustrate the utility of multivariate Granger causality analysis for characterizing large brain networks and introduce a new procedure for removing unimportant nodes while retaining the important ones in the network. This method was applied to tactile perception fMRI data which demonstrated the co-existence of top-down and bottom-up mechanisms in tactile perception.

15:00         3598.     Localization of the Language Network Using Resting State Functional Connectivity MRI and Diffusion Tensor Tracking

Computer 51

Joshua S. Shimony¹, Adrian A. Epstein¹, S Neil Vaishnavi¹, Mark W. Sansbury¹, Jeff G. Ojemann², Abraham Z. Snyder¹, Marcus E. Raichle¹

¹Washington University School of Medicine, St. Louis, Missouri, USA; ²University of Washington School of Medicine, Seattle, Washington, USA

Although the language network is localized around the left sylvian fissure in a great majority of the population, there is a significant amount of variability in localization between individuals.  Localization in individuals is of great interest for pre-surgical planning to help decrease the morbidity from left temporal lobe surgery.  The purpose of this project was to study the variability of the language network in individuals and to validate its localization by correlating the results using a functional method, resting state functional connectivity MRI, and an anatomic method, diffusion tensor tracking.

13:30         3599.     Discrepencies Between Functional Connectivity Measured with BOLD and CBF in Major Depressive Disorder

Computer 52

Richard Cameron Craddock¹, Christopher B. Glielmi², Paul E. Holtzheimer³, Xiaoping P. Hu², Helen S. Mayberg

¹Georgia Tech, Atlanta, Georgia, USA; ²Georgia Tech and Emory University, Atlanta, Georgia, USA; ³Emory University, Atlanta, USA

Functional connectivity was compared between CBF and BOLD for networks associated with depression.  Overall there was consensus with striking differences.

14:00         3600.     Spatially Independent Component Analysis for Automatically Delineating Brain Functional Connectivity with Resting-State FcMRI

Computer 52

Quan Zhu¹, Chung-Yi Yang², Eldom Toh³, John H. Gilmore³, Weili Lin³

¹Duke University, Durham, North Carolina, USA; ²National Taiwan University Hospital, Taiwan; ³University of North Carolina at Chapel Hill, Chapel Hill, USA

In this work, automatic procedures were developed to determine the components that most likely reflect brain functional connectivity for resting-state fMRI images using spatial independent component analysis (sICA).   The developed approaches were then utilized to depict cortical connectivity in neonates (2-4wks), one-year and two-year old children.

14:30         3601.     Detecting Direct and Indirect Functional Connections Using Granger Causality

Computer 52

Zhenyu Zhou¹,², Yonghong Chen¹, Guojun He¹, Paul Wright¹, Mingzhou Ding¹, Yijun Liu¹

¹University of Florida, Gainesville, USA; ²Southeast University, Nanjing, People's Republic of China

An analysis approach that based on the Granger causality method is proposed to study the directional effective influence between brain regions in fMRI data. Granger causality was employed to extracted BOLD signals to investigate effective connectivity in the brain network. Data from an emotion task were used to validate this method, and the emotion pathway was clarified illustrated focus on the right amygdala. The presented results, exploring directional influences between activated brain regions, indicated that the Granger causality mapping combined with  principal component analysis could improve existing applications of traditional pairwise Granger causality mapping method in human brain  connectivity studies.

15:00         3602.     A Comparison of the Abilities of Structural Equation Modeling, Autoregressive Analysis, and Granger Causality to Detect Path Weight Differences in Effect Connectivity Results Calculated from Real FMRI Time Series Data [Not Available]

Computer 52

Suzanne Theisen Witt¹, M Elizabeth Meyerand¹

¹University of Wisconsin, Madison, Wisconsin, USA

Structural equation modeling (SEM), autoregressive analysis (AR), and Granger causality (GC) are three commonly used methods for calculating effective connectivity from fMRI data.  A comparison of these methods’ abilities to detect differences in path weights between two different simple motor tasks is presented.  The results of the comparison suggest that, although the path weights estimated using SEM exhibit a much larger dynamic range than those obtained using AR and GC, these latter two methods may be as good or better than SEM at detecting path weight differences.

13:30         3603.     The Effects of Time Length on Resting-State Functional Connectivity

Computer 53

Jun Xie¹, Gao-Hong Wu¹, Shi-Jiang Li¹

¹Medical college of Wisconsin, Milwaukee, USA

In this study, we investigated the effects of time length on resting-state functional connectivity. EPI data were acquired at 3T in 10 resting normal subjects. The patterns of motor cortex connectivity were compared using different scan time lengths. Our data suggest that about 400 second scanning provides the best tradeoff between the quality of connectivity map and scan time.

14:00         3604.     Spatiotemporal Dynamics of Low Frequency Fluctuations in BOLD FMRI of Rats

Computer 53

Waqas Majeed¹, Matthew Magnuson¹, Shella Keilholz¹

¹Georgia Institute of Technology / Emory University, Atlanta, Georgia, USA

This abstract presents:

1) Spatial patterns evolving over time in low frequency BOLD fluctuations in rat cerebral cortex

2)Preliminary classification of these fluctuations into different 'states'

14:30         3605.     Measuring Connectivity in Sensory Cortex Using an FMRI Adaptation Paradigm

Computer 53

Laura M. Parkes¹, Rishma Vidyasagar¹, Stephen Folger²

¹University of Liverpool, Liverpool, UK; ²Elon University, USA

fMRI adaptation paradigms probe neural activity on a millisecond timescale. We investigate whether this approach can be developed to measure ‘connectivity’ between two regions in the somatosensory cortex. Paired vibrotactile pulses are given to digits 2 and 4 at intervals of 50, 200 and 400ms. A reduction in BOLD amplitude is seen in all subjects at 200ms compared to 50ms and 400ms. This could be due to the first pulse inhibiting neighbouring regions, reducing excitability, and so reducing the response to the second pulse. This demonstrates a method of directly measuring the influence and timing of interactions between sensory regions.

15:00         3606.     The Anatomical Basis of Rest Functional MRI

Computer 53

Aviv Mezer¹, Yossi Yovel¹, ofer Pasternak¹, Tali Gorfine¹, yaniv assaf¹

¹Tel Aviv University, Tel Aviv, Israel

Basal activity measured with fMRI raised tremendous interest among neuroscientists since functional brain activity could be retrieved while the subject rests (resting state fMRI). We show here that the basis of the resting-state fMRI signal – the blood oxygenation level dependent (BOLD) contrast – has a strong anatomical basis in addition to its functional components. We found high similarity between the clusters of the repeated BOLD measures and atlas segmentation. We suggest that multi-dimensional brain MRI acquisition can be used for parcellation of the brain into regions that could not be extracted with conventional MRI.

fMRI: ICA & Clustering Analysis

13:30         3607.     Independent Vector Analysis for Group FMRI Processing

Computer 54

Jong-Hwan Lee¹, Matthew Marzelli¹, Ferenc A. Jolesz¹, Seung-Schik Yoo¹

¹Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts, USA

We propose to develop an independent vector analysis (IVA) scheme to address the permutation problem of the conventional ICA-based group processing scheme. IVA operates to increase independence across output vector components while maintaining dependence among scalar elements within each output vector component (i.e. across frequency bins within the same output index). The ¡®dependence¡¯ in group fMRI processing is analogous to mutual/similar activation patterns across subjects, which are comparable to the group trend in activation.

14:00         3608.     A Method for Group Difference Enhancement by Constraining Mixing Coefficients of ICA Framework

Computer 54

Jing Sui¹, Vince Calhoun¹,²

¹The Mind Research Network, Albuquerque, USA; ²University of New Mexico, Albuquerque, USA

We proposed a novel algorithm called CCICA (constrained coefficient ICA) by incorporating prior statistical group information as constraint into the general ICA framework. Our algorithm improved the capability for identifying the independent component showing the largest group difference related to the mixing coefficients. Applications to both hybrid and real fMRI data demonstrate that CCICA is a promising method for identification of components that can best distinguish schizophrenia patients and healthy controls, and further, it may be more sensitive for extracting features that can serve as potential brain imaging biomarkers of disease.

14:30         3609.     Assessing Reliability of ICA Estimates by Resampling and Clustering

Computer 54

Radu Mutihac¹

¹University of Bucharest, Bucharest, Romania

Analysis of artificially generated fMRI-like data drawn from a single-shot gradient-echo MR EPI block-based visual stimulation paradigm was performed by the stochastic neuromorphic extended Infomax algorithm implementing spatial independent component analysis (sICA). Reliability assessment of the results was carried out by resampling-based techniques and clustering. Fourier and wavelet transforms, which contain weak implicit models of temporal noise, and whitening, which is driven by an explicit noise model, were employed to resampling and setting the appropriate thresholds. Variance estimate that is highly correlated with the separation error was used as model selection criterion for optimizing the parameters of the sICA algorithm.

15:00         3610.     fMRI Clustering Based on Connectivity Profiles

Computer 54

samuel emeriau¹,², Fabien Giersky³, Laurent Pierot³, Eric Bittar¹

¹Université de Reims Champagne-Ardennes, Reims, France; ²Philips Systèmes Médicaux, Suresnes, France; ³University Hospital of Reims, Reims, France

fMRI clustering based on connectivity profiles offers two major advantages in comparison with the classical clustering method: it allows dealing with the problem of spatial correlation of noise that can lead to bad mergings in functional domain and it allows defining a new functional dimension exclusively based on the data without taking the paradigm into account trough specified regressors. The resulting clusters form a partition of the data in homogeneous regions according to both spatial and functional connectivity points of view.

fMRI: Mechanisms

13:30         3611.     Investigating the Source of BOLD Nonlinearity

Computer 53

Nanyin Zhang¹, Xiao-Hong Zhu¹, Wei Chen¹

¹University of Minnesota, Minneapolis, Minnesota, USA

The nonlinear effects in BOLD signal have posed a serious problem in rapid ER fMRI designs. To solve this problem, detailed investigation of the source of BOLD nonlinearity is essential. However, little effort has been spent in separating the potential source of neuronal activity attributing to the nonlinearity observed in BOLD data. The obscurity in this issue seriously impedes accurate modeling of BOLD nonlinearity. Therefore, to find the originality of the BOLD nonlinearity, we have employed a paired-stimulus paradigm in absence of neuronal refractoriness and measured the corresponding BOLD responses. The results indicate that the BOLD nonlinearity can completely be vascular origin.

14:00         3612.     Transient and Steady-State Components of the FMRI BOLD Signal in Somatosensory Cortex

Computer 53

Michael Marxen¹, Tara L. Dawson¹, Fred Tam¹, Simon J. Graham¹,²

¹Baycrest Centre for Geriatric Care, Toronto, Canada; ²University of Toronto, Toronto, Canada

Previous MEG studies have demonstrated transient and steady-state evoked reponses to vibrational touch stimuli to a finger with spatially separated dipole locations in primary somatosensory cortex. The goal of this study is to extract transient and steady state components from the fMRI BOLD response to repeated vibrational stimuli of different durations. In the first three subjects of an ongoing larger study, we found differences in the transient and steady-state activation maps that may reflect transiently and continuously firing neuronal populations. A paired comparison with MEG transient and steady-state dipole locations is in progress.

14:30         3613.     Steady-State to Transient Change of CMR_O2: Dynamic Calibrated FMRI at 11.7T

Computer 53

Basavaraju Ganganna Sanganahalli¹, Peter Herman¹, Fahmeed Hyder¹

¹Yale University, New Haven, Connecticut, USA

We studied the relationship between fMRI BOLD, evoked CBF and the electrical activity of neuronal origin. The BOLD, CBF, CBV and LFP were examined in experiments where the number of the stimuli (2mA, 0.3ms) at rat forepaw was increased from one to four with varying inter pulse interval (IPI) of (167ms, 333ms, 666ms). We found variable dependencies of each measured component (BOLD, CBF, CBV, LFP) with IPI. We found that CMR _O2 contributed to BOLD even with IPI of less than 200 ms (i.e., 6 Hz). These results suggest that BOLD at high field (of 11.7T) has sensitivity to reveal a significant CMRO2 component for extremely short events

15:00         3614.     Assessment of the Significance of Temporal Delays in the BOLD Signal Response to a CO2 Stimulus

Computer 53

julien poublanc¹, Jay Shou Han², Danny Michael Mandell¹, Jeffrey A. Stainsby³, Joseph A. Fisher⁴, David J. Mikulis¹, Adrian P. Crawley¹

¹Toronto Western Hospital, Toronto, Canada; ²Toronto General Hospital, Toronto, Canada; ³GE Healthcare, Toronto, Canada; ⁴Toronto General Hospital , Toronto, Canada

Cerebrovascular reactivity imaging combines BOLD MRI with inhaled CO2 to measure brain’s autoregulatory capacity.  Although the magnitude of the BOLD signal is primarily employed to assess reactivity, the temporal delay of the BOLD response contains useful information, investigated in this study.  We calculated a time delay parameter that significantly correlates with the reactivity parameter, as well as with the time to minimum of the dynamic susceptibility contrast images.  We have also observed that the main time delay  occurs not on onset of hypercapnia but arises from a late return to baseline after the ETpCO2 has already decreased.

13:30         3615.     Quantifying the FMRI BOLD Signal to Characterize the Event-Related Electrophysiological Response

Computer 54

Zhongming Liu¹, Bin He¹

¹University of Minnesota, Minneapolis, Minnesota, USA

In the present theoretical study, we establish a linkage between the quantified BOLD effect size and  the event-related electrophysiological response. Based upon an assumed linear neurovascular coupling, the BOLD effect size at each voxel can be explicitly interpreted as proportional to the time integral of the power of the event-related local synaptic current flow. This relationship contributes to a more rigorous theoretical framework for fusing fMRI and EEG/MEG, which leads to an advanced high-resolution multimodal neuroimaging technique.

14:00         3616.     Towards a Quantitative Relationship Between the BOLD Signal and Deoxyhemoglobin Measured by Near-Infrared Spectroscopy

Computer 54

Dharmesh R. Tailor, MD, PhD¹, Yu Chen, PhD, Ravinder Reddy, PhD

¹University of Pennsylvania, Philadelphia, Pennsylvania, USA

A quantitative relationship is established between the gradient-echo and the spin-echo based BOLD signal and the deoxyhemoglobin concentration measured by near-infrared spectroscopy in the rat brain. This approach provides a more direct means of calibrating the BOLD signal using the recently proposed hyperoxia approach. Furthermore, spin-echo based methods are shown to be sensitive not only to dissolved oxygen but also to deoxyhemoglobin directly. Additionally, spin-echo based T1?-weighted imaging in demonstrated in humans to yield detectable signal changes due to a change in fraction of inspired oxygen.

14:30         3617.     Baseline Cerebral Blood Flow Modulates Functional Perfusion Activation Maps But Not BOLD Activation Maps

Computer 54

Joy Liau¹, Joanna Perthen¹, Thomas T. Liu¹

¹University of California at San Diego, La Jolla, California , USA

This study examines the dependence of functional perfusion or cerebral blood flow (CBF) and BOLD activation maps on baseline CBF.  We used a caffeine dose (200mg) to reduce the baseline CBF and found concurrent decreases in the activation map area based on the CBF response but not the BOLD response.  Also, measures of baseline CBF accounted for a significant portion of the inter-subject variability in CBF activation map area. Factors that can modulate baseline CBF, such as age, medication, and disease, should therefore be carefully considered in the interpretation of studies that use functional CBF activation maps.

15:00         3618.     Strong Correlation of Spin-Echo BOLD Signal with Neuronal Activity in Rat Cortex During Forepaw Stimulation

Computer 54

Ikuhiro Kida¹, Toru Yamamoto¹

¹Hokkaido University, Sapporo, Japan

The relation between signal in neuroimaging techniques and the neuronal activity is essential to investigate brain function. However, the correlation of the BOLD signals with neuronal activity is still unclear. We investigated whether the spin-echo (SE) BOLD signal correlates with neuronal activities during rat forepaw stimulation with various stimulus currents and frequencies (0.5-2.0 mA and 1-10 Hz). We found that the SE-BOLD signal linearly increases with the integration of somatosensory evoked potential amplitude for all stimulus conditions. This result indicates that the SE-BOLD signals may elucidate the neural activity quantitatively.

fMRI: Multimodal & Single Trial

13:30         3619.     Retrospective Synchronization (Resync) Avoids the Residual MRI Gradient Artefact in EEG-MRI Experiments

Computer 55

Hendrik Mandelkow¹, Pascal Halder², Daniel Brandeis², Peter Boesiger¹

¹University and ETH Zurich, Zurich, Switzerland; ²University of Zurich, Zurich, Switzerland

We present a new software synchronization method which improves the post- processing of EEG-fMRI data for the purpose of removing the notorious MRI gradient artefact. Furthermore, we propose a new procedure for quantifying and comparing the cleaning performance of different post-processing algorithms for EEG-fMRI data. The comparison based on in-vivo data as well as simulations of the MGA at high bandwidth shows that the retrospective synchronisation algorithm can improve or even replace hardware synchronisation as well as other post- processing methods such as slice timing correction and PCA. 

14:00         3620.     Comparison of SQUID and MR Detection for Weak Magnetic Fields

Computer 55

Martyn Paley¹, Eugeny Krjukov¹, LiSze Chow¹

¹University of Sheffield, Sheffield, UK

This study compares the sensitivity of a single channel SQUID and a 3T MRI system using phantoms and investigates axonal bursts in the median nerve using transcutaneous electrical nerve stimulation (TENS) stimulation using the SQUID.

14:30         3621.     Simultaneous FMRI/CBV and EEG During Spike-Wave Seizures in WAG/Rij Rats

Computer 55

Asht Mangal Mishra¹, Damien John Ellens, Ulrich Schridde,¹², Joshua Ethan Motelow¹, Michael J. Purcaro¹, Fahmeed Hyder¹, Hal Blumenfeld¹

¹Yale University School of Medicine, New Haven, Connecticut, USA; ²YMPI for Biological Cybernetics, Tübingen, Germany

In absence and other generalized seizure disorders, an abnormal electrical activity termed as spike-wave discharges (SWD) is observed by EEG. fMRI during SWD in humans and animal absence epilepsy models show both increases and decreases in BOLD signals. Here we show regional BOLD increases/decreases during SWD accompanied by CBV increases/decreases in nearly the same areas for anesthetized WAG/Rij rats. Further work will be needed to determine if regions of BOLD decreases during SWD represent vascular steal, a primary neuronal mechanism, or a defect in neurovascular coupling. These findings will be important for the interpretation of various fMRI signals in human absence epilepsy.

15:00         3622.     Simultaneous Single Unit and BOLD FMRI Recordings

Computer 55

D Aksenov¹, Limin Li¹, G Iordanescu¹, Alice M. Wyrwicz¹

¹ENH Research Institute, Evanston, Illinois, USA

Brain hemodynamic functional signal and the corresponding neuronal activity are recorded in an awake, behaving rabbit at high fields.

13:30         3623.     Detection of Single-Trial Events in BOLD FMRI Without Prior Stimulus Information

Computer 56

Cesar Caballero¹, Natalia Petridou¹, Sue Francis¹, Ian Dryden¹, Li Bai¹, Penny Gowland¹

¹University of Nottingham, Nottingham, UK

The aim of this work is to map in space and time the brain’s response to single stimuli without prior knowledge of stimulus timing. Detection of single-trial events in BOLD fMRI time series is a challenge especially in the presence of physiological and systematic fluctuations. Here, a procedure based on Wiener deconvolution is proposed to detect single-trial events with no prior information on stimuli timing. Automatic detection is carried out only assuming a general shape for the hemodynamic response function, and employing tissue-dependent noise characterization. This technique allowed identification of individual trials, and the spatiotemporal evolution of the BOLD response.

14:00         3624.     Increased Statistical Power of in Event-Related Real-Time FMRI (ErfMRI) Using Individual Hemodynamic Response Functions: First Results at 3T and 7T

Computer 56

Maurice Hollmann¹, Tobias Moench¹, Sebastian Baecke¹, Michael Luchtmann¹, Claus Tempelmann¹, Joerg Stadler², Johannes Bernarding¹

¹University of Magdeburg, Magdeburg, Germany; ²Leibniz Institute for Neurobiology, Magdeburg, Germany

Assumptions about the hemodynamic response function (HRF) are used as basis in many fMRI analysis methods. However, the variability of the HRF between subjects, and intra-subject between different brain regions may reduce the statistical power if this assumptions are improper. Acquiring more data to counteract this effect is generally not feasible in real-time fMRI. We analyzed the functional data in event-related real-time fMRI with individual region-based HRF estimates acquired during the experiment-session prior to the main measurement. Using individual HRF estimates instead of the SPM canonical HRF led to an improvement in the statistical significance by an average of 25%.

14:30         3625.     Effects of High Field MR Scanner on Simultaneous EEG Data Quality for Single-Trial Discrimination

Computer 56

Cheng-Yu Wei¹, Robin I. Goldman¹, Paul Sajda¹, Truman R. Brown¹

¹Columbia University, New York, New York, USA

The emerging technology of recording simultaneous EEG/fMRI allows direct measuring and comparison of brain responses to stimuli. Being able to follow changes in attention, adaptation or habituation on a single trial basis by using single trial EEG data to reveal brain regions which correlate with variations in these should considerably advance our understanding of cognitive processing. The simultaneously recorded EEG at 3T, examined with power spectrum analysis were of high quality to allow these single-trial investigations. Quality Assurance tests confirmed that fMRI data, in terms of spatial and temporal SNR, was acceptably stable across time to carry out brain functional studies.

15:00         3626.     A Model Phantom for Investigating Concurrent EEG/fMRI

Computer 56

Gerða Björk Geirsdóttir¹, Matthew J. Brookes¹, Karen J. Mullinger¹, Winston X. Yan¹, Peter Morris¹, Richard W. Bowtell¹

¹University of Nottingham, Nottingham, UK

Simultaneous EEG/fMRI is a technique that makes functional imaging with high spatiotemporal resolution possible.  However, artefacts induced in EEG recordings by the MR scanner make this challenging. The use of an EEG phantom to conduct simultaneous EEG/fMRI experiments provides a means for validating and improving artefact correction techniques, and for investigating possible sources of the observed artefacts.  Here, two different types of saline loaded agar phantoms are described, a dipole phantom which enables measurements of signals with a specific time-course, and a flow phantom with an added aqueduct designed specifically to mimic blood flow.

Bone

14:00         3627.     Advantages of Isotropic Voxel Size for Classification of Trabecular Bone Struts and Plates in Micro-MR Images

Computer 56

Branimir Vasilic¹, Jeremy Magland¹, Michael Wald¹, Felix W. Wehrli¹

¹University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA

A new method, based on local inertial anisotropy, has recently been introduced to classify trabecular elements into rods and plates. It relies on a locally calculated tensor of inertia to estimate the class (rod/plate) to which each voxel in the image belongs as well as the orientation of the corresponding rod or plate. While the structure of trabecular bone is anisotropic and can be used to justify the use of high-resolution imaging with an anisotropic voxel size, we show here that images acquired with isotropic voxel size have advantages in faithfully representing the underlying structure of trabecular bone.

14:30         3628.     On-Line Prospective Registration of Trabecular Bone MR Images for Longitudinal Examinations

Computer 56

Chamith S. Rajapakse¹, Jeremy Magland¹, Felix W. Wehrli¹

¹University of Pennsylvania School of Medicine, Philadelphia, USA

In micro-MRI studies of trabecular bone designed to evaluate structural changes in response to therapeutic intervention, follow-up scan volumes do not typically align with the baseline data due to inaccuracies in patient positioning and scan prescription. These limitations can be overcome by incorporating prospective registration into the acquisition protocol. The technique is based on registering 3D localizer images, acquired prior to the high-resolution images, on-site with an algorithm relying on maximizing the correlation between baseline and follow-up localizers. In phantom and human subject studies it is shown that prospective registration yields very close matching between baseline and follow-up imaging volumes.

15:00         3629.     Proton NMR Study of Transverse Relaxation of Rabbit and Rat Cortical Bone

Computer 56

Henry H. Ong¹, Suzanne L. Wehrli², Felix W. Wehrli¹

¹University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA; ²NMR Core Facility, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA

The NMR signal from cortical bone results from protons in collagen as well as water in the Haversian system. Characterizing transverse relaxation behavior may provide insight into bone composition and architecture. Here, we calculated T2* and T2 components of rabbit and rat cortical bone specimens from bi-exponential fits of FIDs and CPMG signal decays. Both species showed a similar short T2* component arising from collagen. The rat showed a larger short T2 fraction, suggesting a smaller fraction of Haversian canals compared with rabbit. The majority of the bone signal is comprised of water with T2* and T2 values of 200-400us.

15:30         3630.     Density, Structure and Texture Quantitation of Bone Trabeculae at 3.0 Tesla

Computer 56

Gary Paul Liney¹, Clare P. Bernard¹, Chris M. Langton, Lindsay W. Turnbull¹

¹University of Hull, Hull, UK

1) To validate the use of bone volume fraction (BVF) derived from MRI as a surrogate measure of volumetric density. 2) To investigate the structural and textural relationships of bone trabeculae. 3) To compare imaging coils and sequences in terms of their ability to demonstrate these relationships in clinically relevant protocols, and to obtain an optimised protocol for in vivo acquisitions.

14:00         3631.     A Distance Weighted Directional Gradient Method for Fully Automatic Bone Segmentation of Knee MRI

Computer 57

Feng Huang¹, Xiaoming Chen², Deliang Ye¹, Sarah Hertel³

¹Invivo Corporation, Gainesville, Florida, USA; ²University of Florida, Gainesville, Florida, USA; ³Invivo Corporation, Pewaukee, Wisconsin, USA

Fully automatic knee segmentation is desired since it is both time- and cost-efficient. The widely adopted Chan-Vese (CV) model could be used to implement a fully automatic segmentation technique. However, direct application of CV model on knee segmentation has a difficulty in separating different bones (Femur, Tibia and Patella) since these bones have similar intensity level. In this study, a distance weighted directional gradient method is proposed to automatically separate these bones, and define masks for further bone segmentation. With this technique, different bones can be isolated and fully automatic bone segmentation can be achieved without using any prior shape information.  The proposed method was tested on 14 data sets of knee MRI. Results show that the proposed location detection technique can always provide accurate bone locations.

14:30         3632.     Accuracy in Trabecular Bone Structural Measurement from High-Resolution Images Acquired with GRAPPA-Based Parallel Imaging

Computer 57

Ben Hyun¹, Suchandrima Banerjee¹,², Sharmila Majumdar¹,²

¹University of California, San Francisco, San Francisco, California , USA; ²University of California, Berkeley, Berkeley, California , USA

High-resolution magnetic resonance images of the distal radius were acquired to assess the accuracy of bone micro-architecture measurements using GRAPPA-based parallel imaging. Unaccelerated and accelerated images were acquired then analyzed using 2D histomorphometric based analysis to assess image fidelity and reproducibility.

15:00         3633.     Comparison of Mechanical Properties of Trabecular Bone Derived from μMRI and μCT

Computer 57

Chamith S. Rajapakse¹, Jeremy Magland¹, Suzanne L. Wehrli, X Henry Zhang², X Sherry Liu², X Ed Guo², Felix W. Wehrli¹

¹University of Pennsyslvania School of Medicine, Philadelphia, USA; ²Columbia University, New York, USA

Studies have shown that prediction of bone strength can be improved by including structural properties in the analysis in addition to bone density. The aim of this work was to address to what extent mechanical properties of trabecular bone (TB), measured in terms of Young’s and shear moduli, derived from _μMRI, compare with those obtained by _μCT at high resolution. To answer this question TB samples were imaged using _μMRI and _μCT. The data indicate that mechanical constants correlate well between the two modalities and the correlation can be improved by correcting for the bone volume fraction mismatch between them.

15:30         3634.     Investigation of Subchondral Bone Cysts and Cartilage Defects Using in vivo 9.4T MRI in a Model of Osteoarthritis

Computer 57

David D. McErlain¹,², Joe S. Gati,², Vasek Pitelka,², Jeffrey Mason,², Rob Bartha,², David W. Holdsworth,²

¹Robarts Research Institute, London, Canada; ²The University of Western Ontario, London, Canada

The main purpose of this study was to use high-field MRI, combined with micro-computed tomography, to assess the various bone and cartilage abnormalities associated with a rodent model of Osteoarthritis.  The use of pre-clinical models allows for the testing of future pharmacological interventions to treat Osteoarthritis.  The benefit of using this technology allows us to scan live rats repeatedly, over 4 week intervals, in less than an hour per session.

From these images, we have quantified the loss of cartilage within the knee joint, as well as the appearance of subchondral bone cysts, after the rat undergoes surgical destabilization.

Spine

14:00         3635.     Comparison of Lumbar Artery Anatomy and Degenerative Disc Disease in Healthy Controls and Patients with Low Back Pain Combination of Dynamic, Contrast Enhanced Perfusion and MR-Angiogram in the Lumbar Spine Using a 3Tesla MRI

Computer 58

Shadfar Bahri¹, L. Tugan Muftuler¹, Hon J. Yu¹, Min- Ying Su¹, Jamshid Tehranzadeh¹, Anton Hasso¹, Orhan Nalcioglu¹, Vance Gardner²

¹University of California, Irvine, USA; ²Orthopaedic Education and Research Institute of Southern California, Orange, USA

This study investigates the feasibility of a lumbar MRA and vertebral body-perfusion in a single examination. Since the lumbar arterial anatomy has not fully been investigated in vivo so far, the role of the segmental arteries in perfusion of the vertebral body and ultimately delivery of nutrients to the avascular disc is of interest. 31 Healthy controls and patients with back pain have been scanned and the segmental arteries and discs were graded. The control group (8) all showed healthy discs and normal arterial anatomy. The patient group (23) showed 22/23 (94.7%) unhealthy discs and 13/23 (56.5%) stenotic or occluded segmental arteries.

14:30         3636.     Apparent Diffusion Coefficient and Fractional Anisotropy in the Vertebral Bone Marrow [Not Available]

Computer 58

Yu Ueda¹, Tosiaki Miyati¹, Naoki Ohno¹, Yuko Motono¹, Asuka Kouhara¹, Masaki Hara², Yuta Shibamoto², Harumasa Kasai², Miyuki Asahi²

¹Kanazawa University, Kanazawa, Japan; ²Nagoya City University Hospital, Nagoya, Japan

To assess the state of cancellous tissue, we analyzed ADC and fractional anisotropy (FA) in the vertebral bone marrow using single shot diffusion echo planar imaging. A strong negative correlation was found between ADC and DXA-BMD below moderate marrow fat fraction. Significant correlation was noted between ADC and fat fraction. There was a positive correlation between FA and DXA-BMD, and no correlation between FA and FF in the vertebral bone marrow. In conclusion, diffusion analyses with ADC and FA make it possible to obtain more detailed information of the structure of cancellous tissue and the bone metabolism.

15:00         3637.     Mapping Bone Marrow Composition in the Lumbar Spine at 3.0 Tesla

Computer 58

Gary Paul Liney¹, Clare P. Bernard¹, Chris M. Langton, David J. Manton¹, Lindsay W. Turnbull¹

¹University of Hull, Hull, UK

1) To compare data from two quantitative fat imaging methodologies acquired in the lumbar spine. 2) To map the inter-vertebral and age related variations in fat fraction at high spatial resolution. 3) To assess the relationship between volumetric density and marrow content.

15:30         3638.     MR Spectroscopy in Intervertebral Disc -- A Feasibility Study

Computer 58

Jin Zuo¹, Azucena Rodriguez¹, Xiaojuan Li¹, Thomas Link¹, Jeffrey C. Lotz¹, Sharmila Majumdar¹

¹Univ. of California, San Francisco, San Francisco, California , USA

Intervertebral disc degeneration usually begins with biochemical changes within the disc. Quantifying the concentration of the metabolites in the discs would provide objective and important information that is associated with disc degeneration. In this study, the non-invasive single voxel MR spectroscopy was used to assess the concentration of metabolites in intervertebral disc on cadaveric discs. The peak height ratios of the metabolite was correlated with the Pfirrmann grading scheme. In addition, a preliminary in vivo human study was conducted.

Ultra-Short TE MRI in Musculoskeletal Imaging

13:30         3639.     Ultrashort TE (UTE) Imaging with Off-Resonance Saturation: Creating High Contrast for Short T2 Tissues

Computer 56

Jiang Du¹, Atsushi Takahashi², Christine B. Chung¹, Graeme M. Bydder¹

¹University of California, San Diego, San Diego, California , USA; ²GE Healthcare Technologies, Menlo Park, California , USA

The human body contains a variety of short T2 tissues, including cortical and trabecular bone, tendon, ligaments, etc. Magnetization from these tissues can not be spatially encoded between excitation and acquisition before the signal has completely decayed. These short T2 tissues have a much broader absorption lineshape than the long T2 tissues, making them more sensitive to an appropriately placed off-resonance irradiation. Here we present a technique which combines ultrashort TE (UTE) acquisition with magnetization transfer effect to directly image short T2 tissues with high signal and contrast on a clinical 3T scanner.

14:00         3640.     Quantitative Characterization of the Connective Tissues of the Fingers in Cadaveric Specimens: T1 and T2* Measurements Using Ultrashort Echo Time (UTE) MR Imaging in 3T

Computer 56

Berna Dirim¹,², Jiang Du¹, Sheronda Statum¹, Richard Znamirowski¹, Byung Pak¹, Graeme Bydder¹, Christine B. Chung¹

¹University of California San Diego, San Diego, California , USA; ²Izmir Ataturk Egitim ve Arastirma Hastanesi , Izmir, Turkey

Little or zero signal is detected from the short T2 tissues with routinely used pulse sequences. UTE pulse sequences can detect signals from these tissues; allow them to be imaged and quantified. We present UTE imaging (minimal TE = 8 &[micro]s)and quantitative T1 and T2* characterization of the connective tissues. Six frozen hand specimens were used. Dual echo UTE pulse sequence, for T1 measurement a UTE saturation recovery technique and T2* measurement UTE acquisition with variable TE delays were applied. UTE sequences provide high signal imaging of them and T1, T2* measurements to evaluate them using a clinical scanner.

14:30         3641.     Ultrashort TE (UTE) Imaging of the Extensor Tendon Functional Entheses of the Finger

Computer 56

Richard Znamirowski¹, Mark Bydder¹, Christine B. Chun