Brain Mapping seminar abstracts 2008-09

Saad Jbabdi, Ph.D.
FMRIB Centre, University of Oxford
Thursday, 09/11/08, noon, MGH bldg 149

Bayesian Methods in Diffusion-Weighted MRI
Recent methodological advances have enabled Diffusion-weighted MRI to step from a -rather clever- contrast mechanism, to answering fundamental questions about tissue microstructure and white matter connections in the brain. Amongst the most important breakthroughs in dMRI methods have been achieved using Bayesian modelling. For instance, explicit models of local axon orientations in white matter allow us to characterise the uncertainty inherent to such measurements, and propagate it throughout the brain as uncertainty on fibre pathways. This "probabilistic tractography" method has created a potential for a new "contrast mechanism", whereby regions may be segregated according to their global connectivities rather than local properties. Again, Bayesian modelling may be used to efficiently segregate brain regions on the basis of their connections, whilst objectively estimating the number of sub-regions, using e.g. Dirichlet Process mixture models. Finally, explicit models of brain pathways allow us to test for the existence of a connection using Bayesian model comparisons. This talk will give an overview of the concepts and applications of Bayesian models in diffusion weighted MRI.

Krish Sathian, M.D., Ph.D.
Professor of Neurology, Rehabilitation Medicine and Psychology, Emory University
Medical Director, Rehabilitation R&D Center of Excellence, Atlanta VAMC
Friday, 09/12/08, noon, MGH bldg 149

Multisensory Processing of Vision and Touch
It is now well established that extrastriate visual cortical areas specialized for processing particular aspects of vision are also engaged in processing the corresponding aspects of touch. Thus, the lateral occipital complex is active during both visual and haptic perception of shape, and dorsal frontoparietal areas during both visual and haptic perception of location. Does this cross-modal involvement reflect visual imagery or multisensory representations? Visual imagery appears to play a greater role when haptically perceived objects are familiar, compared to when they are unfamiliar. Together with the viewpoint-independence of cross-modal object recognition, these findings suggest the existence of a high-level, modality-independent representation of shape.

Roger Tootell, Ph.D.
Associate Professor in Radiology at Harvard Medical School
Wednesday, 09/17/08, noon, MGH bldg 149

FMRI Reveals a Special Role for Contrast Polarity in FFA, the Cortical ‘Face Area’
Previous fMRI studies of the ventral stream (the ‘object processing’ regions, in inferior temporal lobe) of human visual cortex have described multiple areas that apparently respond to specific categories of visual stimuli. The best-studied of these regions is the Fusiform Face Area (‘FFA’), which apparently responds selectively to images of faces. Here we instead tested the effects of lower-level visual dimensions, using faces as a common stimulus set, to see if this area responds in a way that is truly ‘face selective’.

We found that fMRI responses in FFA to five of these stimulus dimensions were exactly the same as found at the very earliest area of visual cortex (primary visual cortex, V1). In these respects, FFA is NOT selective for processing faces. However in a single dimension (contrast polarity), the FFA responses were qualitatively distinct in FFA, and distinct from those in V1 and other areas of visual cortex. Specifically, responses to normal polarity faces were greater in FFA, compared to faces of reversed polarity. This supports a role for FFA in face recognition, since familiar faces are very difficult to recognize when contrast is reversed (as in photographic negatives).

Sarang Joshi, Ph.D.
Associate Professor of Bioengineering, University of Utah
Wednesday, 09/24/08, 4pm, MIT bldg 46

Statistics of Shape: Simple Statistics on Interesting Spaces
A primary goal of computational anatomy is the statistical analysis of anatomical variability. A natural question that arises is how does one define the image of an "average anatomy". Such an "average" must represent the intrinsic geometric anatomical variability present. Large deformation diffeomorphic transformations have been shown to accommodate the geometric variability but performing statistics of diffeomorphic transformations remains a challenge.

In this lecture I will further extend this notion of averaging for studying change of anatomy on average from a cross sectional study of growth. Regression analysis is a powerful tool for the study of changes in a dependent variable as a function of an independent regressor variable, and in particular it is applicable to the study of anatomical growth and shape change. When the underlying process can be modeled by parameters in a Euclidean space, classical regression techniques are applicable and have been studied extensively. However, recent work suggests that attempts to describe anatomical shapes using flat Euclidean spaces undermines our ability to represent natural biological variability. In this lecture I will further develop a method for regression analysis of general, manifold-valued data.

Further more I will extend the notion of robust estimation to manifold valued data. The median is a classic robust estimator of centrality for data. In this lecture I will formulate the geometric median of data on a Riemannian manifold as the minimizer of the sum of geodesic distances to the data points. I will exemplify the robustness of the estimation technique by applying the procedure to various manifolds commonly used in the analysis of medical images. Using this approach, we also present a robust brain atlas estimation technique based on the geometric median in the space of deformable images.

Christopher Moore, Ph.D.
Assistant Professor of Neuroscience, Massachusetts Institute of Technology
Wednesday, 10/01/08, noon, MGH bldg 149

The Hemo-Neural Hypothesis: A Novel Mechanism for Neuromodulation
Changes in blood flow and volume are well localized to regions of ongoing neural activity, and serve as the basis for fMRI and several forms of intrinsic optical imaging. We have recently proposed (Moore and Cao, 2008) that this process of 'functional hyperemia' may, through several mechanisms, impact neural activity--that in addition to neuro-vascular coupling, there is reciprocal hemo-neural coupling. In this talk, I will describe the motivation for this hypothesis, several mechanisms we anticipate may mediate the translation of signals from the vasculature to neurons, and our preliminary data testing this idea, including novel forms of vascular regulation we are developing. I will also talk about the potential impact of this novel form of neuro-regulation on information processing and/or homeostatic maintenance of cortical dynamics.

References:
The Hemo-Neural Hypothesis: On The Role of Blood Flow in Information Processing, Moore and Cao, J Neurophysiol. 2008; 99: 2035-2047

Wynne Schiffer, Ph.D.
Brookhaven National Laboratory
Wednesday, 10/08/08, noon, MGH bldg 149

Behavioral Imaging in Animal Models of Drug Abuse and Imaging Nanoparticles with PET
I will present two separate lines of investigation that I hope someday will converge, but for now are stronger when pulled apart. Since both lines of investigation would benefit equally from input from investigators at the Martinos Center, my goal is to present two short, discrete talks. The first will cover our progress and problems imaging freely-moving, behaving rodents (with substance use issues). The second will cover a series of PET studies intended to develop kinetic models that can describe (and predict) pathways taken by engineered nanoparticles in living systems.

Behavioral Imaging in Animal Models of Drug Abuse
Preclinical imaging is a rapidly emerging field, providing non-invasive and quantitative measures of fundamental biological processes in living animal subjects. Multi-modal imaging strategies which employ PET, MRI, optical and histological measures can be used to better understand the biology underlying animal models of drug dependence and recovery. This talk will recount our experience with adding a behavioral dimension to these kinds of preclinical imaging studies. Drawing on the success of brain mapping approaches in human subjects, we have developed methods to enable automatic, objective analysis of functional imaging data from behaving rodents. We have successfully used such strategies to determine the pattern of metabolic activations when animals are exposed to environmental cues associated with drug use. Using a new behavioral imaging paradigm, we have shown that environmental cue exposure also increases synaptic dopamine, and it does so to the same degree as cocaine itself. Just as in human imaging studies that measure cue-induced craving or drug-induced high, individual measures of animal behavior (cue-induced craving or cocaine-induced locomotion) correlate with imaging measures of brain function. Taken together, these data highlight a significant role for individual behavioral responses in the interpretation of animal imaging experiments.

Imaging Nanoparticles with PET
Preclinical imaging stands to reap great benefits from nanotechnology, which promises a new generation of radiotracers, contrast agents, molecular targets and imaging instruments. However right now, preclinical imaging can offer nanotechnology precisely what it needs to move forward on all fronts; non-invasive, real-time assays of nanoparticle kinetics and dynamics in living systems. With this information, kinetic models can be derived to describe and predict in vivo nanoparticle fate. Multiple isotope attachments per nanoparticle increase the sensitivity of PET and offer the possibility of tracking, in real time and in space, a single or a few nanoparticles in fully intact mice. Dual labeling approaches using carbon-11 and carbon-14 provide short and long term kinetic measurements from the same animals. The short half-life of carbon-11 permits serial studies of particles that differ only by physical properties such as size (2 vs. 10 nm) or core material (cadmium vs. gold), in the same animal, same scanning session. Serial imaging paradigms can also be used to study the biological responses to nanoparticles, where the first C11-nanoparticle scan is followed by a series of scans to monitor metabolic or inflammatory responses. A host of interesting technological and scientific issues have risen from these studies, from how best to characterize and recover particles to which studies might provide the most useful regulatory information. After presenting a few studies, I will discuss how I have marshaled a diverse array of tools around BNL to resolve some issues, along with those that remain unresolved.

Ken Kwong, Ph.D.
Associate Professor in Radiology at Harvard Medical School
Wednesday, 10/15/08, noon, MGH bldg 149

One EPI Beep, Eight Slices! Applications in Brain Activation and Resting State MRI
We have extended Simultaneous Echo Refocusing echo planar imaging (SER-EPI), a single shot two-slice MR sequence, to acquire up to eight slices (SER-8). We used SER-8 successfully for two applications: 1) to reduce respiratory and cardiac noise in brain activation studies with short TR and cardiac-gated acquisitions; 2) to analyze brain's resting state data acquired with short TR using fuzzy cluster analysis. In resting state studies, we demonstrated the correlation of pixels at the visual and motor cortices, not only at low frequencies (below 0.15Hz), but also at respiratory and cardiac frequencies. We also showed examples of low frequency correlation of local pixels in a water phantom, a response attributed to the MR hardware/scanning processes and dependent on the MR gradient settings.

Joshua Roffman, M.D.
Clinical Fellow in Psychiatry at Harvard Medical School
Wednesday, 10/22/08, noon, MGH bldg 149

Using fMRI to Amplify Subtle Genetic Components of Schizophrenia
Like many other neuropsychiatric illnesses, schizophrenia is strongly heritable, and a more complete understanding of how risk genes contribute to pathophysiology could lead to new and better treatments. However, in most cases, individual risk genes exert very weak effects on clinical symptoms in schizophrenia, which occur far downstream of the genes' biochemical actions. Hundreds of subjects are therefore needed to power clinical gene association studies that focus on single risk alleles, and thousands are often needed to study additive genetic effects. We and others have begun to examine gene effects at the level of brain function, where their signal is presumably stronger, to augment clinical and behavioral findings. ?Many such studies have focused on a functional variant of the COMT gene, which contributes to prefrontal dopamine signaling and thus working memory function. These relatively small studies have produced remarkably convergent findings, attesting to the robust effects of COMT on prefrontal physiology. At the same time, the findings are not diagnostically specific, as similar effects of COMT genotype have been observed in patients, their unaffected siblings, and healthy individuals with no family history of schizophrenia. We have recently found that a functional variant in the MTHFR gene, which has previously been associated with a weak but consistently significant increase in schizophrenia risk, strongly differentiates brain activation during working memory in healthy control subjects and schizophrenia patients, regardless of medication status. Further, when effects of COMT genotype were viewed through the lens of MTHFR, diagnostically specific patterns of brain activation emerged, in a manner consistent with a well elaborated model of prefrontal dopamine signaling. The interactive effects of diagnosis and COMT and MTHFR genotype accounted for 20% of the total variance in left prefrontal activation during the working memory task. These findings could have important implications in view of ongoing clinical trials of folate and dopamine agonists to improve cognition in schizophrenia, and more broadly, they suggest that functional neuroimaging can be used to detect complex genetic effects on brain function using cohorts that are at least an order of magnitude smaller than those required for clinical genetic studies.

Qingzhong Liu, Ph.D.
New Mexico Tech
Monday, 10/27/08, 11:30am, MGH bldg 149

Feature Mining and Pattern Recognition in Steganalysis and Bioinformatics
This talk includes two sections: steganalysis and microarray data analysis.

Steganography is a serious concern of information security, due to the lack of highly accurate detection techniques and the unknown extent to which it is being used by criminals, terrorists, etc., for secretive communications. The objective of steganalysis is to discover the presence of hidden data. In image steganalysis, a Generalized Gaussian Distribution (GGD) model is introduced, the shape parameter is employed to measure the image complexity and to enhance the evaluation of the steganalysis performance. Feature design is discussed to improve the detection of LSB matching steganography. In audio steganalysis, based on the temporal derivatives, a Fourier Spectrum Steganalysis (FSS) is designed and an improved Markov approach is introduced.

Microarray data have high dimension of variables and small sample size, the gene selection is very important to obtain reliable and good analysis. To improve the classification of microarray data, a method called Recursive Feature Addition (RFA) is designed for gene selection. Based on supervised learning and the similarity measures between the candidates and the chosen genes, RFA goes towards the feature set with the best training accuracy and the least redundancy. Experimental results show that it outperforms other compared well-known methods.

Christos Davatzikos, Ph.D.
Professor of Radiology and Director of the Section of Biomedical Image Analysis,University of Pennsylvania
Wednesday, 10/29/08, 4pm, MIT bldg 46

Quantification of Spatial Patterns of Brain Structure and Function
Quantifying spatial and temporal patterns of brain structure and function is a fundamental problem in many neuropsychiatric, neurodegenerative and neurodevelopmental disorders. It is also a very challenging problem, as subtle imaging characteristics must be extracted and integrated into quantities that have diagnostic and prognostic value. We describe our work in the field of computational neuroanatomy, emphasizing the measurement of spatial patterns of brain atrophy and their classification using machine learning tools. We present applications in clinical neuroimaging studies of Alzheimer’s disease and schizophrenia, as well as in measuring brain development in mouse models.

Jelena Kovacevic, Ph.D.
Professor of Biomedical Engineering and Electrical and Computer Engineering, Carnegie Mellon University
Wednesday, 11/05/08, noon, MGH bldg 149

Problems in Biological Imaging: Opportunities for Signal Processing
The question I would like to help answer is: What is the role and what can imaging do for systems biology?

In recent years, the focus in biological sciences has shifted from understanding single parts of larger systems, sort of vertical approach, to understanding complex systems at the cellular and molecular levels, horizontal approach. Thus the revolution of "omics" projects, genomics and now proteomics. Understanding complexity of biological systems is a task that requires acquisition, analysis and sharing of huge databases, and in particular, high-dimensional databases. Processing such huge amount of bioimages visually by biologists is inefficient, time-consuming and error-prone. Therefore, we would like to move towards automated, efficient and robust processing of such bioimage data sets. Moreover, some information hidden in the images may not be readily visually available. Thus, we do not only help humans by using sophisticated algorithms for faster and more efficient processing but also because new knowledge is generated through use of such algorithms.

The ultimate dream is to have distributed yet integrated large bioimage databases which would allow researchers to upload their data, have it processed, share the data, download data as well as platform-optimized code, etc, and all this in a common format. To achieve this goal, we must draw upon a whole host of sophisticated tools from signal processing, machine learning and scientific computing. I will address some of these issues in this presentation, especially those where signal processing expertise can play a significant role.

Wolfgang Wein, Ph.D.
Siemens Corporate Research
Thursday, 11/06/08, noon, MGH bldg 149

Multimodal Ultrasound Fusion for Guidance and Monitoring of Minimally-Invasive Procedures
Apart from its widespread diagnostic use, ultrasound is a valuable imaging modality for minimally-invasive interventions. In this context, it is often desired to integrate pre-operative imaging and planning information with live freehand ultrasound during the procedure. This is associated with challenging multi-modal registration and real-time image processing problems. I will give a detailed description of our technology for image-based CT-ultrasound registration, and present our approach for real-time respiratory motion compensation based on ultrasound. I will furthermore introduce a number of GPU-accelerated algorithms for efficient simulation and reconstruction of ultrasound. All of that is put into context by describing clinical applications in interventional oncology and cardiology.

Souheil Inati, Ph.D.
Assistant Professor of Neural Science and Psychology, New York University
Wednesday, 11/12/08, noon, MGH bldg 149

Reducing Susceptibility Artifacts in FMRI Using Multi-Echo EPI with Dynamic B0 and R2* Correction
The differences in the magnetic susceptibilities of tissue, bone, and air lead to variations in the static magnetic field (B0) and increases in the signal decay rate (R2*). The resulting image artifacts present one of the biggest technical challenges to high field FMRI. We have recently developed a set of data acquisition and image reconstruction tools to address this problem, and have demonstrated their effectiveness in several FMRI studies at 3T. The method combines a multi-echo gradient echo EPI pulse sequence with a non-linear algorithm to simultaneously estimate images of the spin density, B0, and R2* at each time point in the FMRI scan. Our method removes nearly all of the geometric distortions and recovers a substantial portion of the signal lost in standard single-shot EPI with conventional Fourier image reconstruction and is robust to subject motion.

Leo Grady, Ph.D.
Senior Research Scientist, Siemens Corporate Research
Wednesday, 12/03/08, 4pm, MIT bldg 46

Image Segmentation and Compressed Sensing with Random Walks
An ideal image segmentation algorithm could be applied to the segmentation of objects (e.g., tissue classes) without any adjustment for image acquisition device or application. However, a general-purpose, multiway segmentation of objects in an image/volume remains a challenging problem. In this talk, I will describe a recently developed approach to this problem this has been successful in several Siemens products. This segmentation approach inputs a few labeled points from a user (e.g., from mouse clicks) and produces a segmentation by computing the probabilities that a random walker leaving unlabeled pixels/voxels will first strike the labeled set. These probabilities may be computed analytically and deterministically by noting the exact mathematical equivalence with a combinatorial Laplace equation. The solution of the combinatorial Laplace equation admits interpretation of the algorithm as a steady-state electrical circuit simulation or as a minimization of the Dirichlet energy. Going beyond the segmentation problem, we may employ this energy minimization interpretation to minimize the p-norm of the spatial gradient of any function over neighboring pixels. By setting p=0 (i.e., maximizing sparsity), I will show how we can also apply this energy minimization algorithm to image reconstruction for compressed sensing. Although compressed sensing and image segmentation appear to be very different problems, the conclusion of this work is that a common energy minimization approach may used to produce very good results for both problems.

Yoshio Okada, Ph.D.
Professor of Neurology and Neurosciences, Director, Biomedical Research and Integrative Neuro- (BRaIN) Imaging Center, University of New Mexico, Albuquerque, NM; UNM Partner Site Director, Mind Research Network, Albuquerque, NM
Wednesday, 12/17/08, noon, MGH bldg 149

MEG in Human Brain Development Research
I will review the use of MEG in studying brain functions in human fetus in utero and in newborns, infants and children. Fetal MEG is useful because of the vernix, which insulates the fetus electrically, but does not attenuate MEG signals. MEG is useful for measuring spontaneous activity. Evoked responses are weaker, but they may become useful. MEG studies of newborns are in the area of somatosensation, audition and cognition. I will first review a number of interesting studies with adult MEG systems applied to newborns and infants. Then, I will describe the performance of the babySQUID I, a prototype system optimized for infant studies. We have shown that an MEG system optimized for babies provides high sensitivity and spatial resolution. I will describe the use of babySQUID I to study cerebral palsy, epilepsy and autism. I will conclude by describing advantages we anticipate for the babySQUID II, a whole-head MEG system optimized for babies.

Myles Jones, Ph.D.
The Centre for Signal Processing in Neuroimaging and Systems Neuroscience (SPINSN), Department of Psychology, University of Sheffield
Wednesday, 01/14/09, noon, MGH bldg 149

Stimulus-Evoked Neurovascular and Metabolic Coupling Relationships in Rodent Somatosensory Cortex
Blood oxygen level dependent (BOLD) functional magnetic resonance imaging (fMRI) relies on changes in cerebral blood flow, blood volume and blood oxygenation (collectively referred to as the hemodynamic response) that accompany increases in evoked neural activity produced by stimulus presentation or task. As such, data from such techniques can be difficult to correctly interpret in terms of the underlying cerebral activity. As in the MGH laboratory, at Sheffield, optical techniques (LDF, optical imaging spectroscopy) have been used to characterise the cortical hemodynamic response function in rodents following presentation of sensory stimuli. Combinations of optical, MRI and multi-depth electrophysiological techniques have allowed investigation of the quantitative relationships between evoked activity and hemodynamics that underpin fMRI. To attempt to clarify the role of the hemodynamic response in terms of oxygen delivery to brain, the changes in cortical tissue oxygenation during activation were characterised by making polarographic electrode measurements. While recording from superficial cortical locations (1- ~600 µm) ‘electrical’ whisker-pad stimuli typically evoked decreases in brain tissue oxygenation while at deeper cortical locations (~700-1400µm) increases in oxygenation were observed. Similar cortical ‘depth-profiles’ of tissue oxygenation changes could be observed following presentation of stimuli to individual vibrissa. Finally, while recordings were made at superficial depths (~300µm) the spatial extent of cortical activation was varied by either presenting stimuli to the individual whisker (C1) corresponding to the topographic placement of the electrode or to ‘neighbouring’ rows of mystacial vibrissae on the whisker pad (rows D and E). Stimuli presented to whisker C1 again produced decreases in tissue oxygenation while the evoked tissue oxygen response to ‘neighbouring-rows’ stimuli was characterised by increases in cortical tissue oxygen preceded by a smaller initial decrease. These data suggest that the relative degrees of cortical ‘hyper-’ and deoxygenation observed during cortical activation may depend on both the spatial extent of activation and cortical depth location of the recording electrode.

Blaine Chronik, Ph.D.
Associate Professor of Physics and Astronomy, University of Western Ontario
Wednesday, 01/21/09, noon, MGH bldg 149

Variable Fields in MRI: Opportunities and Challenges
The ability to alter the strength of the main magnetic field as a function of time is one of the last system variables yet to be taken full advantage of in MRI. In this talk we will outline ongoing efforts in our labs to explore the ability to vary field strength in MRI. We have been developing variable field MRI to address the problems of combined MRI/PET systems and field-cycled MRI. We have also been pursuing the development of a new technology for obtaining specificity in magnetic resonance molecular imaging. This technology utilizes an additionally electromagnet which is used to modify the strength of the main magnetic field as a function of time in an otherwise standard MRI scanner. Due to the unique response in relaxivity of activated contrast agents, this technology allows one to specifically identify the location of activated contrast agents within an MR image for the first time. We call this new invention "dreMR", which is an acronym for: "delta relaxation enhanced magnetic resonance".

Thomas Liu, Ph.D.
Associate Professor of Radiology and Bioengineering, University of California, San Diego
Wednesday, 01/28/09, 4pm, MIT bldg 46

The Role of the Baseline Physiological State in Functional MRI
Since its inception over a decade ago, most functional magnetic resonance imaging (fMRI) studies have relied upon the blood oxygenation level dependent (BOLD) signal as a measure of neural activity. However, because the BOLD signal reflects the complex interplay of neural, vascular, and metabolic processes, such a straightforward interpretation is not always valid. Indeed, there is growing evidence that modulations of the baseline physiological state can greatly alter the BOLD signal, in some cases independent of any changes in neural activity. Taking this effect into consideration is especially important for fMRI studies of clinical populations, where factors such as disease and medication can significantly modulate the baseline state. In this talk I will review the ways in which the baseline physiological state can modulate the BOLD signal and discuss the impact of this modulation on the interpretation of the BOLD signal.

Maria Angela Franceschini, Ph.D.
Assistant Professor in Radiology at Harvard Medical School
Wednesday, 02/04/09, noon, MGH bldg 149

Neurovascular coupling revisited: secondary synaptic activity and not thalamic inputs drives hemodynamic response
In a series of experiments in humans and small animals we investigated the relationship between the electrical and vascular signals measured during neuronal activity. We used macroscopic electrical (MEG in humans, EEG in rats) and vascular measurements (diffuse optical imaging (DOI)). In different experiments we modulated neuronal activity with parametric electrical stimuli (humans and rats) as well as with cortical infusion of GABA and use of different anesthetics (rats). The results indicate that the secondary postsynaptic activity in superficial cortical layers—not the initial activity in layer IV—drives the hemodynamic response.

Bruce Jenkins, Ph.D.
Director of Neurochemical Imaging at Martinos Center
Wednesday, 02/11/09, noon, MGH bldg 149

Dopaminergic regulation of hemodynamics in the brain with applications to reward and neurodegeneration
A myriad of molecules act to regulate homeostasis of blood flow in the brain. A subset of these molecules also act as neurotransmitters, including serotonin, acetylcholine and dopamine. We will present a unified framework for interpretation of dopaminergically mediated changes in hemodynamics in the brain based upon dopamine receptor sub-type agonism and antagonism. We will show how dopamine release in the brain can act upon specific dopamine receptor sub-types on microvesssels and astrocytes to mediate both vasodilation as well as vasoconstriction. These data naturally lead to a framework where negative and positive BOLD or CBF changes can be interpreted based upon selective receptor sub-type stimulation. We will demonstrate how, in favorable circumstances, one may actually calculate meaningful receptor parameters from MRI hemodynamic data. We will further demonstrate how one can begin to dissect out vascular from neuronal coupling of brain activity to hemodynamic changes. Lastly, we will demonstrate the application of pharmacologically related signal changes to models of drug screening, drug abuse and neurological/psychiatric diseases.

Margaret Lentz, Ph.D.
Instructor at Harvard Medical School
Wednesday, 02/18/09, noon, MGH bldg 149

HIV and the Brain: Early Detection of Neuronal Injury and the role of Antiretroviral Therapy
The success of antiretroviral therapy has reduced the incidence of severe neurological complication resulting from human immunodeficiency virus (HIV) infection. However, increased patient survival has been associated with an increased prevalence of protracted forms of HIV encephalitis leading to moderate cognitive impairment. Thus, neuroAIDS remains a great challenge to patients, and the need for early markers of neuronal dysfunction is imperative. The development of preclinical models and translational studies suitable for testing promising new compounds with neurotrophic and neuroprotective capabilities is of critical importance. A summary of recent SIV and HIV studies from this lab which combine MR spectroscopy, pathology and immunologic results will be discussed, including the impact of antiretroviral therapy on the brain and emerging pathogenesis models.

Mohammed R Milad, Ph.D.
Assistant Professor of Psychiatry at Harvard Medical School
Wednesday, 02/25/09, 4PM, MIT bldg 46

Translational Research in the Neuroscience of Fear Extinction: from rats to healthy humans to psychopathology
Some people adapt well in the aftermath of traumatic events and are quickly able to inhibit their fear responses to trauma-associated stimuli. Fear responses, however, persist for longer periods of time for others to the point where they reach a pathological state. Why are some people more resilient to trauma while others are not? What are the neural substrates that underlie fear inhibition and extinction? Are these circuits deficient in patients with anxiety disorders? In my talk, I will focus on presenting translational data from the rat and human brains with the objective of trying to provide some preliminary answers to the above stated questions. Specifically, I will review human studies indicating that prefrontal areas homologous to those critical for extinction in rats. Furthermore, I will present some data to show that those brain regions in the rat brain appear to be structurally and functionally homologous to specific brain regions in the human brain. I will also show some data suggesting that these brain regions, the ventromedial prefrontal cortex (vmPFC) and the dorsal anterior cingulate cortex (dACC), appear to be deficient in patients with posttraumatic stress disorder (PTSD). I will present some structural and functional neuroimaging and psychophysiological studies done in our lab that focused on the neural mechanisms of fear extinction, particularly extinction recall and the contextual modulation of extinction recall. These recent studies suggest that: 1) human vmPFC is involved in the recall of extinction memory; 2) the size of the vmPFC might explain individual differences in the ability to modulate fear among humans; 3) hippocampal activation is observed during the recall of extinction memory in a context where extinction training took place but not in the initial conditioning context; 4) and the dACC may be involved in the expression of fear responses. I will also present recent neuroimaging and psychophysiological data from PTSD patients suggesting that 1) the retention of extinction memory is impaired in PTSD, and 2) the function of the vmPFC and dACC (measured by fMRI) appears to be impaired in PTSD in the context of fear extinction. Implications of these findings to the pathophysiology of anxiety disorders such as PTSD and current extinction-based behavioral therapies for anxiety disorders will be discussed.

Wanmei Ou
Ph.D. candidate, Computer Science and Artificial Intelligence Lab (CSAIL), MIT
Wednesday, 03/04/09, noon, MGH bldg 149

Spatio-Temporal EEG/MEG Inverse Solver
In this talk, I will present a novel L1L2-norm inverse solver for estimating the sources of EEG/MEG signals. Based on the standard L1-norm inverse solvers, this sparse distributed inverse solver integrates theL1-norm spatial model with a temporal model of the source signals in order to avoid unstable activation patterns and ``spiky'' reconstructed signals often produced by the currently used sparse solvers. The joint spatio-temporal model leads to a cost function with an L1L2-norm regularizer whose minimization can be reduced to a convex second-order cone programming (SOCP) problem and efficiently solved using the interior-point method. The efficient computation of the SOCP problem allows us to implement permutation tests for estimating statistical significance of the inverse solution. Validation with simulated and human MEG data shows that the proposed solver yields source time course estimates qualitatively similar to those obtained through dipole fitting, but without the need to specify the number of dipole sources in advance. Furthermore, theL1L2-norm solver achieves fewer false positives and a better representation of the source locations than the conventional L2 minimum-norm estimates. I will also discuss future directions in EEG/MEG source estimation using priors based on other functional imaging modalities.

Edward Boyden, Ph.D
Benesse Career Development Professor at MIT Media Lab
Wednesday, 03/18/09, noon, MGH bldg 149

Optical Neural Control: Towards Systematic Parsing of the Role of Cell Types in Normal and Abnormal Neural Computation
The brain is three-dimensional, densely-wired, and composed of heterogeneous primitives. Understanding how these elements work in real time to mediate behavior and computation, and how they are compromised in pathological states, is a top priority. We have recently revealed methods for real-time optical activation and silencing of specific cell types, using naturally-occurring molecular sensitizers such as channelrhodopsin-2 and halorhodopsin. We also have created optical hardware and algorithms for systematically testing the contribution, at both the levels of necessity and sufficiency, of brain regions, cell types, and circuit connections to behavioral function. These technologies are designed to be scalable and systematic, and to confront the detailed three-dimensional nature of the brain. We discuss the application of these technologies to the analysis of neural dynamics.

Tommi Raij, M.D., Ph.D.
TMS Core Director at Martinos Center
Wednesday, 03/25/09, 4PM, MIT bldg 46

Dynamicity of Brain Functional Localization
Using functional magnetic resonance imaging (fMRI) and magnetoencephalography (MEG) in healthy humans, we show that functional specificity, i.e., that certain stimuli and tasks are processed in specific brain locations, is modulated by stimulus presentation rate. We also discuss mathematical modeling results of our data suggesting that the underlying mechanism is repetition suppression (“adaptation”), and specifically, adaptation lifetime differences across brain hierarchical levels.

Patrick Purdon, Ph.D.
Instructor at Harvard Medical School
Wednesday, 04/01/09, noon, MGH bldg 149

The Neural Mechanisms of General Anesthesia
General anesthesia is a drug-induced, reversible condition comprised of five behavioral states: loss of consciousness, amnesia, analgesia, lack of movement, and hemodynamic stability. Over 100,000 patients annually receive general anesthesia in the United States for surgical and medical procedures, yet the mechanisms for general anesthesia remain a mystery of modern medicine. While considerable safety improvements in anesthetic drugs and monitoring have been made over the past several decades, anesthesia-related morbidities, including post-operative recall, post-operative cognitive dysfunction, and intraoperative cardiovascular and respiratory depression, remain a significant medical problem. In this edition of Brainmap, I will be presenting results from our recent studies of propofol anesthesia in humans using EEG/fMRI, EEG, and intracranial recordings.

Mert Sabuncu, Ph.D
Postdoctoral Associate, MIT CSAIL
Wednesday, 04/08/09, noon, MGH bldg 149

Image-driven Population Analysis through Mixture Modeling
In this talk, I will present *iCluster*, a fast and efficient algorithm that clusters a set of images while co-registering them using a parametrized, nonlinear transformation model. The output of the algorithm is a small number of template images that represent different modes in a population and an image-driven partitioning of the population. This is in contrast with traditional, hypothesis-driven computational anatomy approaches that assume a single template to construct an atlas and partition the population based on clinical/demographic/genetic data. We derive the algorithm using a generative model of an image population as a mixture of deformable template images. I will present experimental results that demonstrate the utility of iCluster in several contexts. In one experiment, I will motivate a multiple atlas strategy to develop automatic segmentation tools for a pool of subjects that consists of healthy controls and schizophrenia patients. Next, I will show how we employed iCluster to partition a data set of 416 whole brain MR volumes of subjects aged 18 through 96 years into three sub-populations, which mainly correspond to age groups. The templates reveal significant structural differences across these age groups that confirm previous findings in aging research. In another experiment, we ran iCluster on a group of dementia patients and healthy controls. The algorithm discovered two modes that correspond to a healthy majority and a sub-population of patients with dementia. These experiments suggest that the proposed image-driven clustering strategy can be used to discover sub-populations associated with interesting structural or functional "modes."

Randy Buckner, Ph.D.
Professor in Psychology, Center for Brain Science at Harvard University; Associate Director at Martinos Center
Wednesday, 04/15/09, noon, MGH bldg 149

The Brain Genomics Superstruct Project
(Superstruct: To build over or upon another structure; to erect upon a foundation)
How genetic factors play a role in brain function has been exceedingly difficult to understand. One challenge is that psychiatric and neurologic illness may be influenced by dozens (or even hundreds) of genes each playing a small role in directing brain development. To understand such complexity, studies of genes and brain function will require thousands of participants. In this talk, I will discuss a novel approach to obtaining brain and genetic data by adding a small additional study to the many ongoing studies of brain function being conducted across the local research community. The approach is possible because of recent advances in imaging technology developed at the Martinos Center that provide methods to image the structure and function of the brain rapidly, in under 15 minutes. Moreover, the imaging methods allow for estimates of basic properties of brain organization such as which hemisphere is dominant for language and the overall level of cortical connectivity. Behavioral data, including timed cognitive tests and personality questionnaires, are obtained via the web. Pilot data in 200 subjects indicates the approach is feasible.

Derrick MacFabe, M.D.
Director, The Kilee Patchell-Evans Autism Research Group, University of Western Ontario, London, Canada
Special time: THURSDAY, 04/16/09, 11AM, MGH bldg 149

Use of Time-of-Flight Secondary Ion Mass Spectroscopy (ToF-SIMS) Imaging as a Tool to Examine Widespread Metabolic Markers in a Novel Rodent Model of Autism
Recent research is examining ASD as a systemic encephalopathic condition involving immune, digestive and metabolic dysfunction exacerbated by environmental triggers in genetically sensitive subpopulations. Clinical observations suggest that certain gut and dietary factors may transiently worsen symptoms in ASD. Furthermore, pre and post natal infectious processes, and antibiotic exposure have been implicated as possible risk factors for ASD. There is a growing need to develop animal models and imaging techniques to simultaneously examine the disparate behavioral and metabolic derangements in ASDs. Time-of-Flight Secondary Ion Mass Spectroscopy (ToF-SIMS) is an extremely powerful technique to image diverse biomolecules on object surfaces, including those of biological tissues. We have been using ToF-SIMS to image widespread ASD related biochemical markers in our novel propionic (PPA) rodent model of autism. Propionic acid (PPA), is a short chain fatty acid, and an important intermediate of cellular metabolism. PPA is also a fermentation by-product of a subpopulation of opportunistic enteric bacteria (i.e clostridia, propionibacteria), a putative risk factor for ASD, and is also a common food preservative. PPA and other short chain fatty acids (i.e butyrate and acetate), affect diverse ASD associated physiological processes such as cell signaling, neurotransmitter synthesis and release, mitochondrial function, lipid metabolism, immune function, and gene expression. Infusions of PPA in rats induce repetitive, antisocial behavior, electrographic changes (seizure, caudate spiking) and neuropathology ( innate neuroinflammation, increased oxidative stress markers, glutathione reduction) consistent with those found in ASDs. Similarly, ToF-SIMS analysis of sections of PPA treated rat brain reveals clear images of forebrain structures (neocortex , external capsule, thalamus, hippocampus) which were grossly similar to control brain. However, ToF-SIMS showed increased Na, Cl, and N derivatives in external capsule, consistent with extracellular edema and increased oxidative stress. PPA reduced signal intensity of cholesterol and phosphatidyl choline in white matter. These findings are consistent with ASD as a white matter disorder of increased oxidative stress, inflammation and altered lipid profiles. Collectively, these central effects of PPA suggest a possible link between diet, seizure, movement disorder, social impairment, neuroinflammation, increased oxidative stress, mitochondrial dysfunction, environmental sensitivity and gastrointestinal symptomotology found in ASD. We propose that some types of ASD may be partial forms of genetically inherited or acquired disorders of altered short chain fatty acid metabolism, resulting in increased exposure to these enteric metabolites at critical times during the life cycle. Tof-SIMS may be an extremely powerful research tool to examine these widespread changes in ASD and other neurological disorders.

Venkatesh Murthy, Ph.D.
Professor of Molecular and Cellular Biology, Harvard University
Wednesday, 04/29/09, noon, MGH bldg 149

Multimodal optical imaging of the mouse olfactory bulb: sensory representation, plasticity and blood flow
Mice use ~1000 odorant receptor types to probe chemical space. Receptor neurons in the nose project their axons to the olfactory bulb and form anatomical units called glomeruli. Odor information is then processed by the postsynaptic circuitry in bulb, which includes feedback projections from cortical areas and brainstem neuromodulatory centers. The modular organization, the superficial location and the sensory nature of the olfactory bulb allow us to study many aspects of brain function at subcellular resolution in living animals. I will present three examples of studies from my laboratory using different forms of optical imaging. First, we used wide-field microscopy and a large panel (~100) of odorants to analyze the nature of odor representation in the glomerular layer. Second, we used wide-field and multiphoton microscopy to reveal the influence of a neuromodulatory system (serotonin) on odor representation in the olfactory bulb. Third, we used optical microscopy to examine the role of astrocyte signaling in the coupling of neural activity to blood flow in the glomerular layer.

Lucia Vaina, M.D., Ph.D., D.Sc.
Professor of Biomedical Engineering, Boston University
Wednesday, 05/06/09, noon, MGH bldg 149

Plasticity and Reorganization in the Human Visual System after Stroke
Research in my laboratory combines psychophysics, computational modeling and structural or functional brain imaging to study the functional architecture of the human visual motion system and its plasticity in the normal and damaged brain. One long-term goal of our studies on visual motion perception is to obtain enough of understanding of the mechanisms involved so that outcomes can be controlled and directed in the clinical setting. To illustrate our approach I will discuss two facets of adult brain plasticity at the systems level, that provide “ecologically” or “cortically” driven solutions to problems created by selective visual motion deficits in stroke patients. First, I will talk about plasticity of sensory-motor routines which I define as exploitation of computational redundancy by employing available, behaviorally relevant, alternate mechanisms and cues. Focusing on tasks of visually guided navigation, I will show how motion impaired patients can cross the ecological gap between their abilities and the demands of the world around them. I will show that when primary cues, such as optic flow, are no longer available, patients may use alternate cues, such as landmarks or scale-change, in order to navigate in their surroundings. Second, in a longitudinal fMRI study I will illustrate plasticity of sensory-motor maps defined as behaviorally driven topographic functional reorganization of cortical areas around the lesion. Combined, these two approaches are a first step in trying to understand how in humans, the motion responsive cortex adjusts to injury, and in trying to identify candidate areas that might mediate visual behavior by using the residual visual mechanisms.

Jacob Hooker, Ph.D.
Research Fellow, Brookhaven National Laboratory
Wednesday, 05/13/09, noon, MGH bldg 149

Development of New Methods for Carbon-11 Labeling and Application to Imaging Tools for the Brain
The incorporation of carbon-11 into small molecules has been paramount to the success of positron emission tomography (PET) for neurological imaging, providing a powerful way to non-invasively study the human brain. Over several decades, the medical imaging group at BNL has used ¹¹C-PET to study drug abuse and human addiction. As part of this effort, we recently labeled salvinorin A, the psychoactive component of S. divinorum, with carbon-11 to study its uniquely fast onset, yet short-lasting hallucinogenic effect. Using salvinorin A as an example, this presentation will focus on how ¹¹C-PET is used to study the brain. Moreover, the fundamental advancements in carbon-11 chemistry that underlie our ability for PET imaging will be discussed, highlighting the need for new chemical methodology. Finally, two new methods for carbon-11 labeling and their application to potential radiotracers will be described.

Sydney S. Cash, M.D., Ph.D.
Assistant in Neurology, MGH and Instructor, Harvard Medical School
Wednesday, 05/20/09, noon, MGH bldg 149

Multiscale Neural Recordings in Human Cortex – Sleep, Cognition and Epilepsy
Recent advances in recording technologies and computer capacity now make it possible to record neural activity at the level of single neurons or small groups of neurons directly from the human cortex. Using a variety of unique microelectrode array technologies in patients undergoing intracranial recordings for localization of seizure onset zones, our group has begun investigating the physiological basis of cognitive processing, sleep components and the pathophysiology of epilepsy from the level of the single neuron to larger scale ensemble activity. As examples, I will present data from microelectrode recordings which show that one element of sleep, the K-complex, is an induced form of a down-state and relate this finding to the cortical physiological which underlies slow-oscillations. The relationship between the patterns of activity seen in these oscillatory phenomenon and yet another element of sleep, the sleep spindle, will be discussed and placed in context of wider scale ensemble events across the cortical surface. These sleep elements will be compared with patterns of activity seen during cognitive tasks.
In contrast to these normally occurring phenomenon, we will also discuss some of the physiological properties of epileptic cortex which have been revealed using combinations of macroelectrode and microelectrode recordings. In particular, we will focus on results indicating new possibilities for delineating epileptogenic cortex from more normal cortex by using recordings at the level of synaptic activity and single neurons. Other, preliminary data will be discussed which point toward novel methods for anticipating and understanding seizures using information from single neurons or small populations of neurons.

Carsten Wolters, Dr.rer.nat.
Research Associate, Institute for Biomagnetism and Biosignalanalysis, Universitätsklinikum Münster, Germany
Wednesday, 06/03/09, noon, MGH bldg 149

EEG/MEG Source Analysis Based on Realistic Finite Element Volume Conductor Modeling and its Application in Presurgical Epilepsy Diagnosis
Bioelectric source reconstruction in the human brain from scalp Electro- and Magneto- encephalography (EEG/MEG) signals is sensitive to head volume conductor properties, i.e., tissue geometries and conductivities. A mathematical dipole is widely used as the model of the primary current source. Since the inverse problem is not uniquely solvable, different classes of approaches exist and will be presented, which are based on different a-priori knowledge. All inverse methods are based on solutions to the corresponding forward problem, i.e., the simulation of the electric potential and the magnetic flux at the head surface for a dipole in the cortex sheet of the human brain.
In my talk, I will propose different approaches to model the dipole singularity introduced into the differential equation within the finite element (FE) method based solution to the EEG/MEG forward problem. The presented approaches will be validated using high-resolution constrain Delaunay tetrahedralizations and geometry-adapted hexahedral meshes in a multi-layer anisotropic sphere model. Methods are discussed to individually estimate the head tissue conductivity parameters. A successful application of the presented approaches to the field of presurgical epilepsy diagnosis will finally be discussed.

Predrag Petrovic, Ph.D.
Karolinska Institutet, Sweden
Wednesday, 06/10/09, noon, MGH bldg 149

Passive and active fear: from man to reptile in the same brain
Anxiety compromises a set of ill-defined affective disorders, including many different and possibly non-related components. In order to better understand these different aspects of anxiety we can related them to basic fear states evident in the evolutionary evolved interaction between prey and predator. Two main fear states that have been described in animals are passive fear (post-encounter state) and active fear (circa strike state). The passive state is important when a predator is sensed in order to avoid disclosure and better attend to the threat. This state is complex and time consuming, and it is mediated by the amygdala in concert with the prefrontal cortex and hypothalamic / brainstem networks. Active fear is necessary when the animal is attacked by a predator, and may be mediated only by the brainstems regions such as the dorsolateral periaqueductal gray (dlPAG) since this state requires a fast and robust response. The present talk will show how these fear states may be studied as models for passive anxiety and panic-attacks in human subjects using functional imaging methods.