Brain Mapping seminar abstracts 2011-12

Archana Venkataraman, PhD
MIT
Archana is a Ph.D. candidate in the department of Electrical Engineering and Computer Science (EECS) at MIT. She works in the Medical Vision Group under the supervision of Prof. Polina Golland. She earned a B.S. in Electrical Engineering from MIT in 2006. In 2007, she completed my M.Eng. in Electrical Engineering at MIT under the supervision of Prof. Alan V. Oppenheim. Her thesis focused on signal approximation using an anti-aliasing frequency warping.

Generative Models of Brain Connectivity for Population Studies
Connectivity analysis focuses on the interaction between brain regions. Such relationships inform us about patterns of neural communication and may heighten our understanding of neurological disorders. Here, we propose a generative framework that uses anatomical and functional connectivity information to find impairments within a clinical population. Anatomical connectivity is measured via Diffusion Weighted Imaging (DWI), and functional connectivity is assessed using resting-state functional Magnetic Resonance Imaging (fMRI). We first develop a probabilistic model to merge information from DWI tractography and resting-state fMRI correlations to infer latent templates of connectivity within the brain. We also present an intuitive extension to population studies and demonstrate that our model learns stable differences between a control and a schizophrenia population. Despite the promise of our joint model, connectivity results are difficult to interpret and validate given our region-centric knowledge of the brain. To alleviate these concerns, we present a novel approach to identify regions, which we call disease foci, associated with the disorder based on connectivity information. This allows us to aggregate pairwise connectivity changes into a region-based representation of the disease. Once again, we use a probabilistic formulation: latent variables specify a template organization of the brain, which we indirectly observe through resting-state fMRI correlations and DWI tractography. The inference algorithm simultaneously identifies both the afflicted regions and the network of aberrant connectivity. Our framework offers a principled method to explore brain connectivity. This allows us to tackle challenging and unexplored problems such as multi-modal analysis and region localization.

Kiho Im, PhD
Boston Children's Hospital
Dr. Im received his PhD in Biomedical Engineering at Hanyang University in Korea, where he investigated sulcal folding patterns, shape, and variability using a surface-based method in adult brains. At the FNNDSC, he has been focusing on the more challenging problem of characterizing sulcal pattern development in fetal brains. He hopes that his analysis tool development will ultimately provide innovative imaging markers for monitoring brain health and development unlike any that have previously been developed.

Individual Gyral Pattern-based Structural Connectivity and Network Analysis: Application to Polymicrogyria Brains
For the structural brain network analysis based on DTI tractography, most previous studies parcellated cortical regions to serve as nodes using volume- or surface-based registration to an atlas. However, an atlas-based node definition at the macroscopic scale appears to be too coarse to characterize short U-fibers connections between adjacent gyri. In addition, typical gyral patterns cannot be identified in some patients with cortical malformation because of abnormal cortical folding shape. We suggested novel node definition and structural connectivity analysis using individual gyral pattern and topology, which is biologically more meaningful and able to provide more explicit description of individual network organization. Using our approach, we investigated the white matter network in polymicrogyria (PMG), which is a cortical malformation with multiple small gyri and abnormal cortical lamination. The connectivity analysis for different fiber groups subdivided based on gyral topology revealed severely reduced connectivity between neighboring gyri (short U-fibers) in PMG, which was highly correlated with the regional involvement and extent of abnormal gyral folding. The patients also showed significantly reduced connectivity between distant gyri (long association fibers) and between the two cortical hemispheres. With relation to these results, gyral node based-graph theoretical analysis revealed significantly altered topological organization of the network (lower clustering and higher modularity) and disrupted network hub architecture in association cortical areas involved in cognitive and language functions in PMG patients. Consequently, our analysis provided more detailed findings and interpretation for the disrupted structural connectivity and network related to the abnormal cortical structure in PMG, overcoming the limitations of an atlas-based approach.

Lauren Atlas, PhD
New York University
Lauren received a PhD in Psychology from Columbia University in 2011, where she completed her doctoral research under the mentorship of Tor Wager. Her research combines human brain imaging, psychophysiological measures, pharmacological manipulations, and behavioral outcomes to understand how expectations shape affective experience. Lauren is currently a post-doctoral fellow at New York University working with Elizabeth Phelps.

Mediation analysis for fMRI: Identifying brain pathways that link experimental manipulations with behavior
Though a great deal is known about how informational cues affect visual perception, far less is known about how they shape affective processes such as pain and pleasure. In this talk I will present recent work examining the brain mechanisms that formally link expectations to dynamic affective experience. I will introduce whole-brain multi-level mediation analysis, a path modeling approach we developed that jointly considers the effects of psychological manipulations on brain responses and the relationship between brain responses and changes in reported experience. The first part of this talk will focus on expectancy effects on pain perception. We used mediation analysis to identify brain regions that 1) are modulated by predictive cues, 2) predict trial-to-trial variations in pain reports, and 3) formally mediate the relationship between cues and reported pain. We found that a subset of "pain matrix" regions mediated cue effects on pain, and that effects on these regions were in turn mediated by cue-evoked anticipatory activity in medial orbitofrontal cortex and ventral striatum, regions not previously directly implicated in nociception. In the second part of this talk, I will present new data identifying brain mediators of the relationship between expectations about healthfulness (i.e. claims of "light" versus "regular" foods) and experienced taste, and discuss the relationship between within- and between-subjects factors in the context of multi-level mediation. Finally, I will address the relationship between causal inference and path modeling approaches, and hypotheses about domain-general versus domain-specific processes.

Jyrki Ahveninen, PhD
MGH Martinos Center
PhD, Psychology, University of Helsinki, Finland

Multimodal Studies on Neuronal Representations of Distance in Human Auditory Cortex
Determining the distance of objects is of key value in many everyday situations. For objects that fall outside the field of vision, hearing is the only sense that provides such information. However, neuronal mechanisms of auditory distance perception are still poorly understood. One of the reasons is that contributions of intensity and distance processing have been difficult to differentiate. Typically, the received intensity increases when sound sources approach us. Nonetheless, we can also distinguish between soft-but-nearby and loud-but-distant sounds, indicating that distance processing can also be based on intensity-independent cues. We combined behavioral experiments, fMRI measurements, and computational analyses to identify neuron populations representing distance, independent of intensity. In a virtual reverberant environment, sound sources were simulated at varying distances (15-100 cm) along the right-side interaural axis. For such stimuli, two intensity-independent depth cues are available: direct-to-reverberant ratio (D/R) and interaural level difference (ILD). Our acoustic analysis suggested that, of these cues, D/R is more reliable and robust than ILD. However, based on our behavioral results, subjects' discrimination performance was more consistent with complex intensity-independent distance representations, combining both available cues, than with representations based on either D/R or ILD individually. fMRI activations to sounds varying in distance (containing all cues, including intensity), as compared to activations to sounds varying in intensity only, were significantly increased in the planum temporale and posterior superior temporal gyrus contralateral to the direction of stimulation. This suggests that neurons in posterior non-primary auditory cortices, in or near the areas processing other auditory spatial features, are sensitive to intensity-independent sound properties relevant for auditory distance perception.

Christos Papadelis, PhD
Boston Children's Hospital
Christos Papadelis is Instructor in Neurology at Harvard Medical School, Research Associate at Boston Children's Hospital, and Manager of the BabyMEG laboratory. He received the diploma in Electrical Engineering from the Aristotle University of Thessaloniki, Greece, in 1998, and his MSc and PhD in Medical Informatics, in 2001 and 2005 respectively, from the same institution.

Tracking Noninvasively the Information Flow along Fibers in the Human Brain: from Accurate Localization of Brain Sources to Functional Connectivity
In order to understand how the brain processes information, we must find out how this information is represented in a distributed fashion among different regions, and how it travels from region to region along neural fibers. Magnetoencephalography (MEG) is the ideal neuroimaging tool to assess the functional dynamics of the human brain activity since it offers both unparalleled temporal resolution and good localization accuracy especially for cortical neural sources. MEG is able to localize accurately even weak superficial sources that generate magnetic activity resembling the human brain high frequency oscillations (Papadelis et al., 2009). The accurate localization of these sources is possible even when just few trials are available, and is more pronounced when special instrumentation is used (Okada et al., 2006). Deeper thalamic sources can be localized with an accuracy of ~2 cm by averaging a high number of trials (Papadelis et al., 2012) in contrast to the traditional belief. The good localization accuracy of MEG let us link the macroscopic observed brain activity with the microscopic cytoarchitectonic identity of the activated regions (Papadelis et al., 2010) in healthy adults. The next step is to assess the functional connectivity among these regions and estimate how the information flows from one region to another. The functional connectivity among different brain regions was assessed in a visuomotor experiment by using well-established techniques from the fields of source localization, cytoarchitectonic anatomy, and multivariate autoregressive signal processing. A large-scale brain network oscillating in the alpha band was found that was active during visuomotor integration involving early visual areas, the dorsolateral prefrontal cortex and the hand motor area. These findings support a new model of how the frontal lobe makes use of visual information to produce conscious choices between competing behaviours, and emphasize the critical role of MEG in studying functional connectivity in the human brain. In more simple tasks, where a very high number of trials can be collected and averaged, the millisecond by millisecond afferent information flow along the thalamocortical fibers can be tracked by using MEG.

Iiro Jaaskelainen, PhD
Brain and Mind Laboratory, Department of Biomedical Engineering and Computational Science, Aalto University School of Science, Espoo, Finland
BIO

Movies as Naturalistic Affective-Social Stimuli in Functional Magnetic Resonance Imaging Studies
Functional magnetic resonance imaging (fMRI), which allows non-invasive measurement of brain hemodynamic activity over timescales of ~seconds, has significantly increased possibilities to study the neural basis of human perceptual and cognitive functions, including emotions and social cognition. Pioneering studies published in 2004 by Bartels et al. and Hasson et al. paved way for the use of highly naturalistic stimuli such as feature films in fMRI studies. While these early studies focused on sensory-cortical processing of visual perceptual features, we showed in 2008 that prefrontal cortical and subcortical (e.g., amygdala) areas involved in emotions and social cognition also exhibit across-subjects replicable responses when watching a contextually primed and emotionally engaging movie clip. In our recent studies, we have utilized movie clips as naturalistic stimuli to probe the cerebral mechanisms underlying emotions and social cognition in healthy volunteers. Results from these studies will be presented in this talk. We have also tested and developed new data analysis methods, including ones that allow estimation of instantaneous (i.e., single-TR) synchrony of hemodynamic activity between subjects and brain areas to study dynamic network activity of the human brain when subjects are watching movies. These freely available data analysis methods will also be briefly introduced in this talk.

Spencer Bowen, PhD
MGH Martinos Center
BIO

Influence of Partial Volume Effect Correction on Kinetic Parameter Estimation for Dynamic Brain PET Imaging
The combination of MR and dynamic PET exams via combined MR-PET may offer significant advantages over standalone imaging for assessing brain function in both healthy volunteers and dementia patients. Loss of quantification in the measurement of tracer uptake for structures with dimensions on the level of the PET spatial resolution (deemed the partial volume effect), however, can significantly bias such studies if not corrected for. Although partial volume effect correction (PVEC) methods have been well characterized for static PET imaging, PVEC on dynamically acquired PET images reconstructed with unregularized iterative algorithms (e.g. OSEM) creates unique challenges that have not been addressed previously. As the reconstructed spatial resolution of PET images may change during the course of a dynamic scan, and accurate PVEC requires an accurate estimate of image resolution, bias may result in the estimated kinetic parameters if this phenomena is not considered. We present several MR based PVEC methods and optimize these techniques for dynamic FDG PET imaging via a combination of studies, including: experimental phantom data, a Monte Carlo simulation of a dynamic brain acquisition using a digital patient phantom, and dynamic MR-PET imaging of healthy volunteers. Results show that kinetic parameter bias and noise are significantly influenced by the choice of the PVEC method, with performance differences between implementations the greatest for structures with the smallest features and most hypometabolic uptake.

Michael Fox, MD, PhD
MGH
Dr. Fox completed his MD/PhD with Marcus Raichle at Washington University in St. Louis followed by Neurology Residency training at the Partners Neurology Program in Boston. He has extensive experience with resting state functional connectivity MRI with over 20 publications, 3,900 citations, and multiple invited lectures and international presentations. He is presently a post-doctoral fellow enrolled in a self-designed clinical and research fellowship focused around brain stimulation. His clinical training is in movement disorders and deep brain stimulation at Partners (Massachusetts General Hospital, Brigham and WomenÕs Hospital) while his research training focuses on noninvasive brain stimulation at Beth Israel Deaconess Medical Center with Dr Alvaro Pascual-Leone. His goal is to use neuroimaging to better understand and improve therapeutic brain stimulation.

Clinical Applications of Resting State Functional Connectivity MRI: A Guiding Light for Brain Stimulation
Resting state functional connectivity MRI (rs-fcMRI) has several theoretical and practical advantages for use in clinical applications and is witnessing an exponential increase in popularity. Although its primary use has been investigating differences that occur in various brain diseases, increasing evidence suggests that it might be used to directly guide therapeutic intervention including neurosurgical resection and brain stimulation. After providing an introduction and overview of these topics, this talk will focus primarily on how rs-fcMRI may be used to guide transcranial magnetic stimulation for diseases such as depression. Potential application in other diseases benefiting from brain stimulation will be discussed.

Ofer Pasternak, PhD
Brigham and Women's Hospital
Dr. Ofer Pasternak is an Instructor in Psychiatry at the Brigham and Women's Hospital, Harvard Medical School. He has a Ph.D. in Computer Science from Tel-Aviv University, Israel, where he developed variational calculus methods for the analysis of diffusion MRI. He was a Fulbright postdoctoral fellowship in the Psychiatry Neuroimaging Laboratory (director, Dr. Martha E. Shenton) and the Laboratory for Mathematics in Imaging (director, Dr. Carl-Fredrik Westin). He was recently awarded the NARSAD young investigator award. His main interest is in the development of imaging tools that help understand the relation between degeneration and neuroinflammation in psychiatric and neurodegenerative brain disorders.

Free-Water Imaging for the Estimation of Extracellular Volume in the Human Brain
Recent diffusion MRI studies have pointed out that eliminating the contribution of extracellular water increases the sensitivity of the diffusion measures to alterations in tissue architecture. Moreover, in brain imaging, estimation of the extracellular volume appears to indicate pathological processes such as atrophy, edema and neuroinflammation. In my talk I will present the free-water imaging method, which assumes a bi-tensor model that accounts for the freely diffusing water molecules that can only be found in the extracellular space. I will discuss various estimation alternatives for the model, and will demonstrate how the method can be used for the study of neurodegenerative brain disorders as well as schizophrenia and traumatic brain injuries.

Jonathan Brooks, PhD
University of Bristol, UK
BIO

Pain: cause, consequence and control
Our understanding of pain has been transformed over recent years. Historically, the pain system was thought to be a relatively simple network conveying nociceptive signals via peripheral and spinal nerves to the thalamus and somatosensory cortex. However, through the use of non-invasive techniques to record brain activity, it is now apparent that a widespread network of cortical areas are involved in creating the sensation we know as pain. The multiplicity of brain regions responding to pain reflects its multi-dimensional nature: there are sensory components that tell us which part of the body has been damaged, we are able to relate how much it hurts by reflecting on previous painful episodes (memory), and we feel how unpleasant the pain is (affect). Furthermore, how much pain we experience depends on factors such as depression, mood, attentiveness, gender, age, coping mechanisms, etc. During this talk I will attempt to describe some of the techniques that have been used to explore pain processing in the brain, brainstem and spinal cord, and introduce some recent developments that are demonstrating how pain, particularly when it becomes chronic, can re-shape the brain. Lastly, I will discuss the principles of endogenous pain control, and show some recent human imaging data highlighting the role of the brainstem and spinal cord in this process.

Sylvain Bouix, PhD
Brigham and Women's Hospital
BIO

Increased Diffusion Anisotropy in Gray Matter in Mild TBI with Persistent Post-Concussive Symptoms
A significant percentage of individuals diagnosed with mild traumatic brain injury experience persistent post-concussive symptoms that do not resolve. Little is known about the pathology of these symptoms and there is often no radiological evidence based on conventional clinical imaging. We aimed to find a physiogenic basis to PPCS following mTBI using DTI. A novel analysis method was developed to identify abnormalities when the location of brain injury is heterogeneous across subjects. The foundation of our method is a normative atlas of diffusion measures built from a population of normal controls across different brain regions of interest.

David Salat, PhD
MGH Martinos Center
BIO

Deterioration of Structural Connectivity in the Aging Brain
A range of work has demonstrated that the neural tissue representing the structural basis of inter-regional connectivity, cerebral white matter (WM), is highly vulnerable to an array of insults as a result of normal and pathologic aging. However, the biological conditions that contribute to such degenerative changes and the cognitive and behavioral consequences of such deterioration are less well understood. Neuroimaging technologies, such as diffusion tensor imaging (DTI) have progressed over the last decades to provide sensitive measures of tissue microstructure that allow for the detailed quantitative assessment of brain anatomy and pathology. Such procedures can provide novel insight into patterns of compromised connective integrity resulting from aging and from age-associated conditions including Alzheimer.s disease (AD) and can contribute to understanding mechanisms that contribute to these conditions. Studies using DTI have provided greater appreciation for spatial patterns of WM deterioration in older adults and individuals with age-associated disease and ongoing work is exploring potential mechanisms by which poor vascular health may interact with risk factors for AD to promote the progression from a mild cognitive syndrome to a debilitating dementia.

Stephen LaConte, PhD
Virginia Tech
BIO

Brain Computer Interfaces Increase Whole-Brain Signal-to-noise
Overt actions allow us to interact directly with our environment. By definition, though, covert mental activity is unobservable by a third party and does not translate to action in the outside world. Real-time functional magnetic resonance imaging (rtfMRI) is a nascent technology that can convert thought into action by transducing noninvasive brain measurements into a control signal to drive physical devices and computer displays, and enable neurofeedback. We have developed an rtfMRI system that is based on multivariate predictive models (e.g. support vector machines) that determine the relationship between the image data and the corresponding sensory/behavioral conditions (brain states). This talk will present recent studies in which we have found that subject-based control involved frontoparietal circuitry and increased the signal-to-noise ratio (SNR) of task-related brain activity. Importantly, the enhanced SNR was highly correlated to improved prediction accuracy of brain state classifiers, and because these classifiers serve as the control signal for neurofeedback, this work suggests the exciting possibility that brain-computer interfaces can be substantially enhanced by taking advantage of this effect.

Mehdi Moghari, PhD
Beth Israel Deaconess Medical Center
Mehdi Hedjazi Moghari received B.Sc. and M.Sc. degrees in Electrical Engineering from Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran in 1999 and 2002. He received Ph.D. degree in Electrical and Computer Engineering at QueenÕs University, Kingston, Ontario, Canada, in 2008. During 2009 and 2012, he continued his research at the cardiac MRI Center of Harvard Medical School at Beth Israel Deaconess Medical Center, Boston, MA, USA. He is now with the Department of Cardiology at Boston ChildrenÕs Hospital. His research interest is on cardiovascular MRI including: pulse sequence design, motion compensation, image registration, and reconstruction.

Cardiovascular Magnetic Resonance Imaging: An Emerging Field in Clinical Practice
Cardiovascular magnetic resonance imaging (MRI) is an emerging modality that provides unique measurements and observations to cardiologists. Although echocardiography, X-ray angiography, and multi-detector computed tomography are conventionally used in clinical practice, cardiac MRI is a noninvasive and complementary technique that offers gold standard measurements for the assessment of regional and global cardiac function, myocardial infarction, and coronary artery disease. To improve the prognostics and diagnostic value of cardiac MRI, images with a high spatial-resolution needs to be acquired. However, the acquisition of high spatial-resolution images is challenging in cardiac MRI due to the respiratory motion of the heart. Prospective compensation of the respiratory motion unpredictably prolongs scan-time that impedes the acquisition of images with a high spatial-resolution. In this talk, the conventional prospective respiratory motion compensation algorithm in cardiac MRI is briefly presented. A new adaptive algorithm is introduced that minimizes the respiratory-induced heart motion in a predictable scan-time. Next, a novel algorithm based on Compressed Sensing (CS) is presented to reduce scan-time while minimizing the respiratory motion of the heart. The reduction in scan-time is then used to improve the spatial-resolution.

Alexandra Golby, MD
Brigham and Women's Hospital
BIO

Multi-modal brain imaging for pre-operative neurosurgical planning and intra-operative guidance
Over the last few decades, intracranial surgery has been transformed from a dangerous, unpredictable intervention with significant morbidity and mortality to a routine elective procedure with most patients leaving the hospital in a few days usually in better or equivalent neurologic condition than they were preoperatively. The change is due to the tremendous advances in imaging, visualization, and operative techniques that allow the surgeon to have a much better understanding of the anatomy and pathology that are the targets of the intervention. Nevertheless, there remain many times when neurosurgeons find themselves uncertain of how to proceed due to a lack of information. An important example is in surgery for primary brain tumors that arise from the brain parenchyma and may variably infiltrate, compress, or destroy brain tissue and that can be very difficult to differentiate from normal brain tissue. All brain surgery needs to consider the functional organization of the brain tissue around the lesion to avoid causing a new neurologic deficit. However, differentiating critical functional areas from areas that can be resected is not possible either on conventional imaging or by inspection at the time of surgery. To decide whether surgery is feasible for a patient with a given lesion, the surgeon requires a complete and accurate map of the complex and critical functional and structural anatomy of that individualÕs brain. A further challenge results from the progressive deformation of anatomy that takes place during the surgical intervention (brain shift), making preoperative images and associated neuronavigation increasingly inaccurate. To address these major issues in modern neurosurgery requires the development of new techniques to demonstrate critical brain anatomy and pathology. Numerous pre-operative and intra-operative techniques are now available which can provide the surgeon with helpful information to guide surgical resection. Advances in neuro-imaging acquisition, computing, and image processing include neuronavigation, intra-operative MRI, multi-modality integration, advanced modeling, and image-analysis. In preoperative planning, functional MRI (fMRI) may be used to locate eloquent cortex and Diffusion tensor imaging (DTI) may be used to demonstrate white matter tracts. Together these methods allow the evaluation of surgical risks, selection of the best method of intervention, and planning of the safest surgical approach. Intraoperative imaging may be used to design the surgical flap, alert the surgeon of surrounding anatomy, define brain shift, assist in more complete lesion resection, and monitor for certain intraoperative complications.