Uri T. Eden, Ph.D. - US grants

DESCRIPTION (provided by applicant): Intellectual Merit: Deep brain stimulation (DBS) is a highly promising therapy for Parkinson's disease (PD). Yet most patients do not get full therapeutic benefit from DBS due to its critical dependence on electrode location, a sweet spot in the dorsolateral posterior sensorimotor subunit of the sub-thalamic nucleus (STN), for therapeutic efficacy. PI Cheng was trained at a center where 70% of DBS patients obtained full therapeutic benefit, improving so markedly that they no longer require any PD medications. Such efficacy is atypical even in academic centers because DBS electrode placement is not standardized, scientific, or systematic. We propose to construct a neural modeling, estimation and control framework for STN, which will enable the development of a new surgical tool that will standardize DBS placement: an automated intraoperative closed-loop DBS localization system. Development of this transformative technology requires: 1) neurophysiologic characterization of the sweet spot. In PD patients, microelectrode recordings will measure single unit spiking activity (action potentials) of STN neurons at different distances from the sweet spot and from within it. Point process models will be estimated from this data and will capture complex stochastic relationships between extrinsic (e.g. behavior) and intrinsic (local neural network activity) factors and STN spiking activity. Principled inferential methods will confirm the sweet spot's existence and characterize its electrophysiological properties; and computational conductance-based modeling will elucidate the ionic mechanisms underlying the sweet spot's physiology. 2) construction of neural estimation and control algorithms for STN DBS. Signal processing and control will derive a robust feature set from STN spiking activity which will reliably predict where the electrode is and will then guide the electrode to the sweet spot. This transformative project requires collaborations between physicians, scientists, mathematicians and engineers with expertise in neurosurgery, neurophysiology, neural signal processing, estimation and modeling, and control theory. For these reasons automation of DBS localization remains largely untapped, giving us the opportunity to lead the scientific development of this next-generation technology. Broader Impact: Due to cost, less than 10% of PD patients worldwide receive DBS. Automating surgical implantation and obviating complex postoperative DBS programming from suboptimal electrode placement would decrease cost, and thus increase patient access. Even greater societal impact, however, would come from improved DBS efficacy, which is life-changing for PD patients. DBS patients of Dr. Cheng have stated that they have been returned to their pre-PD status, and that not just their lives but also the lives of their family members, so long held hostage by a debilitating chronic disease process, have been returned to them. Our proposal attempts to extrapolate these benefits to the larger PD population. Even more importantly, DBS is a nascent procedure holding great promise for many future neurological and psychiatric indications. A technology that improves DBS targeting fidelity and efficacy would hold the potential to improve the lives of millions of patients and their families worldwide. This project will be integrated into curricula in the home and affiliated departments of the PIs. Coursework for signal processing and neuronal spike modeling in the senior undergraduate and graduate levels will gain from our proposal. A graduate level modern control theory course with applications to neural systems will also be developed and offered. Traditional courses in neuroanatomy and neurophysiology will be enhanced by our proposal's insights into the relationships between physiology, anatomy, and function. The PIs also plan to reach out to the academic community by providing representative samples of rare neurophysiological data and analysis code. When cultivated, such a database will provide a platform for investigators around the world to benchmark software algorithms, optimize analog and digital components for new hardware platforms that will process neural signals, and develop a more complete understanding of the mechanisms of DBS. PI Cheng has strong relationships with industry companies including Medtronic, the manufacturer of DBS hardware. We will leverage this to expedite the development and testing of our concept. Our project's outcome may thus have a substantial impact on how DBS systems are designed.

DESCRIPTION (provided by applicant): PROJECT SUMMARY / ABSTRACT Across the world, nearly 50 million people suffer from epilepsy. For an estimated 30% of these individuals, seizures remain poorly controlled despite maximal medical management. Moreover, treatment of epilepsy through medication often results in significant - sometimes debilitating - side effects. To advance the therapeutic management of epilepsy, we must understand the physiological mechanisms which support this disease. Unraveling these mechanisms is especially difficult. Like many neural processes, the seizure involves spatial scales spanning many orders of magnitude, from the individual neuron to the entire nervous system. Moreover, typical clinical recordings provide only a limited view of the vast mechanisms underlying the seizure. In this project, an interdisciplinary research group consisting of clinicians, statisticians, and mathematicians will study the dynamical mechanisms that support epilepsy and, in doing so, suggest novel therapies to treat the disease. The research will focus on three fundamental aspects of the seizure - how it begins, how it spreads over the brain, and how it ends - observed in voltage recordings from human patients with epilepsy. These recordings will span multiple spatial scales and include activity generated by individual neurons, small neural populations, and large brain regions. To characterize these data, sophisticated analysis techniques will be applied that link the activity across spatial scales. In addition, computational models of the neural activity will be developed and constrained using the multiscale data. These models will then be applied to suggest the mechanisms that support multiscale interactions during seizure and to propose novel therapies to treat epilepsy. Completion of the proposed research will represent a significant step forward toward a deeper and more complete understanding of epileptic physiology, and toward a system for exploring and testing innovative methods for seizure control. PUBLIC HEALTH RELEVANCE: PROJECT NARRATIVE Epilepsy is a devastating and poorly understood illness that evolves on multiple spatial scales. During a seizure, multiscale interactions are prominent yet the mechanisms that support these interactions remain unknown. To address (and eventually treat) these mechanisms, an interdisciplinary team of researchers will record, analyze and model multiscale voltage data from human patients with epilepsy.

DESCRIPTION (provided by applicant): The abilities to learn, remember, evaluate and decide are central to who we are and how we structure our lives. These abilities, and indeed the vast majority of cognitive functions, are thought to depend on specific patterns of brain activity. Each new experience is thought to drive a unique pattern of brain activity in the hippocampus, a brain region critical for storing memories for the events of daily life. Subsequent reactivation of this experience after learning is thought to drive a consolidation process that engrains the patterns in hippocampal and cortical circuits. Similarly, subsequent retrieval is thought to rely on the reinstatement of patterns similar to those present during the original experience. Current evidence points to the replay of sequences of hippocampal neurons during sharp-wave ripple events (SWRs) as a candidate mechanism for both memory consolidation and memory retrieval. To determine whether memory replay drives consolidation and retrieval for the associated memory representations, we will carry out directed manipulations that go beyond interrupting all SWRs to target replay events by their content. Our work will build on our expertise in real-time feedback and recent developments in cluster-less decoding that have allowed us to develop all of the technological elements required for real-time, content-based interruption of hippocampal replay events. This will allow us to assess the role of specific memory replay events in memory processes. Our Specific Aims are: 1) to develop an optimal adaptive statistical framework for real-time decoding and interruption of memory replay, 2) to test the hypothesis that hippocampal replay events drive memory consolidation for the replayed memories, and 3) to test the hypothesis that hippocampal replay events support rule learning and the internal exploration of specific future possibilities. Our real-time approach has the potential to create new causal links between the replay of specific patterns of activity and the ability to consolidation memories and to use past experience to guide future decisions.

? DESCRIPTION (provided by applicant): Advances in technology for measuring neuronal activity at ever-larger scales and with increasing spatial and temporal resolution, concomitant with a decrease in costs of data storage, are driving a revolution in neuroscience. The era of big data is reflected in a number of major funding initiatives that have begun worldwide to support neuroscience research, and especially neuronal data collection. As a flood of neuronal data accumulates worldwide, a new challenge faces the global neuroscience community: how to make sense of these complex data to drive basic biological insight and to shed new light on neurological and neuropsychiatric disorders. This new, data-driven era of neuroscientific research demands that investigators master the fundamental methods in time series and image analysis and know when and how to appropriately apply these methods, either in custom applications or in existing software packages. Accessible - yet rigorous - resources to develop hands-on experience with modern data analysis techniques are lacking in neuroscience. To address directly this current and growing worldwide challenge, we propose to develop an innovative open online course (or OOC). To reach the largest target audiences - the biologists, psychologists, and clinicians immersed in neuronal data - we will assume only a basic mathematics background and limited familiarity with computer programming, common to those trained in biological sciences. The proposed OOC will target investigators at all career levels - spanning from the beginning undergraduate researcher to the established PI - to analyze and understand neuronal data. Through an interdisciplinary case-study approach, we will use real-world neurophysiological data (including data available from large, emerging public repositories) to motivate the study of modern quantitative analysis methods. The OOC will comprise 15 independent modules. The first two modules will emphasize programming in MATLAB for neuroscientists and computational techniques relevant for large datasets. Each additional module will focus on one category of neuroscience case-study data, and will consist of multimedia material combining video lectures, MATLAB-based examples, and quantitative assessments. The modular format will provide multiple coherent learning paths through the online content, and thereby allow personalized learning for individuals with varying quantitative backgrounds and research interests. The OOC format will also permit the developed resources to be widely available, disseminated, and discoverable. The proposed OOC will prepare researchers with the fundamental skills required for the analysis of neuronal big data, and elevate the general competencies in data usage and analysis across the research workforce.

? DESCRIPTION (provided by applicant): The brain is a massively interconnected network of regions, each of which contains neural circuits that process information related to combinations of sensory, motor and internal variables. Adaptive behavior requires that these regions communicate: sensory and internal information must be evaluated and used to make a decision, which must then be transformed into a motor output. Despite the importance of this question, we know relatively little about the principles of how spiking activity in one region influences activiy in downstream areas, particularly in the context of cognitive operations like decision-making. Here we propose to address this question by focusing on how the ventral striatum (VS), a region critical for motivational control of behavior receives and processes information from two important upstream regions, the orbitofrontal cortex (OFC) and the hippocampus (HP). We have assembled a unique team of scientists with complementary expertise studying the HP (Frank), OFC and VS (Kepecs), using synergistic technologies for large-scale recordings using novel polymer electrodes (Frank/Tolosa) with improved optogenetic identification of projections (Kepecs), and a team of statistical and computational researchers providing complementary analytical expertise in dimensionality reduction (Machens), statistical modeling (Eden/Kramer) and normative models (Ganguli). Our combined expertise will allow us to (1) measure large populations of neurons across the brain regions, (2) identify and (3) manipulate the neurons connecting them in order to (4) test for the first time a range of hypotheses about different modes and circuits for information transmission across regions. Beyond revealing how the OFC, HP and VS communicate during learning and decision-making, our approach will provide new experimental tools and computational methods for systems neuroscience, as well as new insights into the general principles of information transmission across regions.

PROJECT SUMMARY Although rhythms are a prominent feature of brain activity, the role of rhythms in brain function (and dysfunction) remains elusive. Rhythms have been proposed to organize information transfer within and between brain regions by modulating neural excitability at different time scales. Rhythms have also been proposed to interact across these different time scales, a phenomenon labeled cross-frequency coupling or CFC. Clinical and experimental observations have identified many different types of CFC, such as coupling between the phase of a low frequency rhythm and the amplitude of a high frequency rhythm (phase-amplitude coupling), or between the phases of two different frequency rhythms (phase-phase coupling). Many functional roles for CFC have been proposed, including in working memory, neuronal computation, communication, learning and emotion. Despite the mounting experimental evidence for CFC, three important challenges remain that limit understanding of this phenomenon. First, many different data analysis methods have been developed to characterize CFC, with each method typically focused on one type of CFC. Choosing an inappropriate method weakens statistical power and introduces opportunities for confounding effects. Second, analysis of CFC typically occurs post hoc, prohibiting opportunities to modulate CFC during an experiment. New methods are needed to assess CFC in real time while limiting the impacts of potential confounds. Third, the mechanisms that produce CFC are not known. While computational models developed to explore these mechanisms provide important insights, these models have been mainly restricted to synaptic mechanisms of rhythm generation and associations between two types of rhythms. New models are needed to examine the role of other rhythms and rhythm generating mechanisms in CFC. Inclusion of more realistic biological features in simulations of neural rhythms facilitates exploration of a new challenge: how electrical stimulation modulates CFC. In this project, an interdisciplinary research group consisting of a statistician, a mathematician, and a psychiatrist-engineer will analyze, model, and modulate cross-frequency coupling. To do so, the team will develop and apply a statistical inference framework suitable for real time analysis of CFC, and apply this framework to analyze - and modulate with electrical stimulation - in vivo recordings from rat cortex and subcortex. The team will also develop computational models of CFC, to link the observed data to cellular mechanisms, and create hypotheses testable in the in vivo experiments. Completion of the proposed research will represent a significant step forward toward a more complete understanding of cross-frequency coupling, and toward a system for exploring and testing innovative methods for its modulation.

PROJECT SUMMARY Epilepsy is the world?s most prominent serious brain disorder, affecting nearly 50 million people worldwide. For an estimated 30% of these patients, seizures remain poorly controlled despite maximal medical management, with significant financial costs and effects on health and quality of life. To advance the therapeutic management of epilepsy requires a more detailed understanding of the spatiotemporal dynamics that drive seizures. Characterizing these dynamics is especially difficult because, like many brain functions, the processes span spatial and temporal scales, from the fast activity of small neural populations to the slow evolution from seizure onset to termination of large brain regions. How brain signals at one scale relate to those at other scales is a significant and poorly understood issue. While animal models of epilepsy provide powerful techniques to investigate detailed neural activity within and between spatial scales, the relationship of these models to human epilepsy is unclear. An alternative to animal models of epilepsy is to study spontaneously occurring seizures in vivo from a population of human patients. However, typical in vivo clinical recordings provide only a limited view of a seizure?s multiscale dynamics. In this project, an interdisciplinary research group consisting of epileptologists and clinical neurophysiologists, a statistician, and a mathematician will study the spatiotemporal dynamics of human seizures. To do so, the team will analyze simultaneous microelectrode and macroelectrode recordings from human patients during seizures, with a particular focus on the organized spatiotemporal patterns and high frequency oscillations common in epilepsy. To make sense of these data, the team will develop and apply new methods to characterize these patterns, and link these activities to candidate mechanisms in computational models. Completion of the proposed research will represent significant progress towards a deeper understanding of human seizures, new methods to analyze and model the spatiotemporal dynamics of seizures observed in complex multiscale data, new methods to estimate model parameters and variables from brain voltage recordings, and new candidate targets for surgical treatment of epilepsy.