Nathan E. Crone - US grants

DESCRIPTION(adapted from applicant's abstract): The overall objective of the proposed research is to identify and validate electrophysiological signatures of human cortical processing and to use them to study the neural mechanisms of motor, sensory, arid language functions. Previous EEG and MEG studies have linked regional cortical activation with event-related changes in the energy of alpha (8-13 Hz) and gamma (40 Hz) oscillations. Using electrocorticographic (ECoG) recordings from subdural electrodes implanted for clinical purposes, we have confirmed these observations and discovered an event-related broad-band augmentation of power in high gamma frequencies (80-100 Hz). This novel index of function cortical activation is more discretely localized in space and time, and more consistent with functional-anatomic and neurophysiological (ERP) correlates, than previously described spectral changes. Gamma oscillations have been associated with the synchronization of neuronal assemblies during cortical processing. In contrast, suppression of alpha oscillations is thought to arise from the activation of thalamocortical circuits that otherwise gate cortical processing. We therefore propose to test the following hypotheses: (1) that broad-band "high" gamma ERS indexes task-specific cortical processing, and that (2) alpha desynchronization facilitates this processing via thalamocortical circuits. Our specific aims are directed toward testing the (A) spatial and (B) temporal correspondence between these spectral indices of cortical activation and the following benchmarks: (i) general knowledge of the functional anatomy of perceptual, motor, and language tasks, (ii) specific functional-anatomic information derived from cortical stimulation mapping for clinical purposes, (iii) the latency and duration of perceptual stimuli and motor responses, and (iv) spatiotemporal properties of event-related potentials. The generalizability of our conclusions from these ECoG studies will be tested in a group of normal subjects using high density scalp EEG. Information derived from this research will facilitate future investigations of the neurophysiological correlates of human brain activation, including studies of the timing and topography of normal and disordered cognitive/neuronal operations.

DESCRIPTION (provided by applicant): The overall project goals are to study the cortical network dynamics of human upper limb motor control spanning two distinct spatial scales recorded with electrocorticography (ECoG), and to demonstrate that these dynamics can be estimated in real-time and used to control the JHU Applied Physics Lab Modular Prosthetic Limb (MPL) during execution of functionally useful complex action sequences. Our human subjects will be instructed to perform complete functional movements characteristic of activities of daily living. We will analyze the task-related temporal evolution in the strength and pattern o interactions among large-scale cortical networks known to be recruited in visually-guided reach-to-grasp tasks. Using multi-scale subdural ECoG with combinations of routine clinical macro-electrodes (2.3 mm diameter, 1 cm spacing) recording activity of broadly spread elements/nodes of neural networks, and inset arrays of microelectrodes (75 ?m diameter, 0.9 mm spacing) recording the activity of local sub-networks, we will test our overall hypothesis that there is a functional hierarchy between the two scales (Aim 1). More specifically, we hypothesize that large-scale network dynamics involving premotor/motor cortex reflect the evolution of sensory-motor processing demands during complex action sequences, while micro-scale population activity and network dynamics in motor cortex reflect the low-level kinematics of these tasks. We will utilize methods of estimating dynamic effective connectivity developed by our team to study interactions between these scales and test whether there exists a spatially heterogeneous and hierarchical structure within the macro-micro scale networks. The results of these analyses have wide-ranging clinical implications for both the optimal scale of functional mapping for clinical diagnostic purposes and the extent of implantations for neuroprosthetic control. We will exploit multi-scale ECoG recordings and online estimates of the dynamics of neural activation and large-scale/local network interactions to achieve control of the MPL during functionally useful tasks (Aim 2). This approach will go beyond traditional paradigms that have developed neural control over individual degrees of freedom. We will do this by embedding low-level control within an innovative framework whereby knowledge of task goals supplement direct kinematic decoding. This project will build on our team's previous successes in implementing a system for semi-autonomous ECoG control of the MPL, employing machine vision and route-planning algorithms, during complex interactions with objects requiring the coordination of multiple joints. This system will be able to leverage for the first time the rich complexity of temporally and spatially resolved network dynamics correlated with high-level goals to achieve functionally useful control of an advanced neuroprosthetic limb.

? DESCRIPTION (provided by applicant): To avoid post-operative language impairments after surgery for drug-resistant epilepsy, clinicians rely primarily on electrocortical stimulation mapping (ESM), but this can trigger afterdischarges, clinical seizures, or cause uncomfortable sensations, all of which can prevent mapping in some areas. Moreover, ESM can be very time- consuming and the resulting impairment is usually all-or-none, complicating its interpretation. These practical limitations have long motivated passive electrocorticography (ECoG) as an alternative, or complementary, functional mapping technique that can map function at all sites simultaneously, resulting in significant time savings without adverse side-effects. Recent technical developments also permit ECoG functional mapping to be performed online during testing, but the correspondence between these results and ESM has not been as good for language mapping as it has been for motor mapping. This may be, in part, because even simple language tasks such as object naming or word reading require the recruitment and interaction of widely distributed and potentially redundant cortical areas responsible for different stages of cognitive processing, and because there is no a priori threshold for a magnitude of activation critically important for task performance.. Our overall hypothesis is that the functional importance of a cortical site depends on its role in task-related network dynamics, i.e. the propagation of activation between distributed cortical regions performing the distinct cognitive operations necessary for successful task performance. In this project we will study the task-related network dynamics of spoken word production in order to improve the accuracy of ECoG language maps and to better understand the relationship between ECoG and ESM language maps. First, we will use ECoG to capture the fine temporal dynamics of network propagation during a series of simple word production tasks. We will decompose these network dynamics into temporally cascaded subnets corresponding to functional-anatomical modules responsible for each task's constitutive cognitive operations. In addition, we will identify high centrality noes that serve as hubs facilitating propagation across subnets. Second, we will test these ECoG models of task-specific network dynamics by temporarily deactivating the hubs of high efficiency pathways linking subnets. We will test the effect of this deactivation on verbal latency and accuracy using much briefer (200- 400 ms) stimulation trains than those used during routine clinical ESM (2-5 seconds). This will also test the feasibility of performing ESM with a lower risk of afterdischarges and seizures, and with greater functional specificity. Third, we will compare both ECoG network mapping and ESM to ground-truth post-operative language outcomes in order to comparatively assess their predictive abilities. Although the immediate goal of these studies is to gain deeper insights into the cortical network dynamics of spoken word production and how they are affected by ESM, these studies will exert their most profound and lasting impact by improving the clinical utility of ECoG for both extraoperative and intraoperative functional mapping prior to respective surgery.

Most current approaches to understanding the neural basis of cognitive processes are severely limited in two respects. First, most commonly used methods do not have the temporal (e.g., fMRI) or spatial (e.g., MEG/ EEG) resolution to capture the relevant dynamics. Second, even methods with high spatio-temporal resolution (intracranial EEG - icEEG) typically approach target cognitive processes in a fragmentary, un- integrated way. For instance, language is typically studied as a conglomeration of separate subsystems: perception, pattern recognition, categorization, semantically/syntactically appropriate response selection, cross-modal integration, motor control and sensorimotor integration. The present proposal aims to remedy both limitations by using icEEG to study a model system, reading/speech/language, from an integrative and unified perspective. We focus on reading, a complex task that involves visual pattern recognition, visual- auditory and visuo-motor integration, semantic, syntactic and phonological access, and (in reading aloud) - response selection and motor sequencing. Reading allows for easy, yet ecologically valid manipulations of cognitive load in the language system. The neuro-computational framework we propose to test is that computation is achieved not by information passing through a sequence of discrete processing stages in individual modules but via state transitions of a distributed network. We will recruit a large cohort of 80 patients in whom we will quantify both local as well as inter-regional cortical dynamics during word reading - from early primary visual perception, through selection, to word output. We will leverage our established techniques for precise co-localization and analysis of grouped icEEG data, circumventing the sparse sampling problem inherent to human icEEG experiments. The combined use of sub-dural grid electrodes and stereo-electroencephalographic depth electrodes will enable the study of not only classic peri-sylvian regions, but also of deep sulci (and regions such as the planum temporale). We will then characterize dynamic network interactions using linear and non linear measures of amplitude covariance in high frequencies, following analyses we have developed previously. Critical nodes and critical transitions in network states will then be perturbed using closed-loop activity-triggered direct cortical stimulation. To achieve these goals we have set up a collaboration between the Texas Comprehensive Epilepsy Program and Johns Hopkins Medical Center - both centers have a proven record of studying language with icEEG. Our team has expertise in all aspects language, reading, icEEG signal analysis, population level network modeling from intracranial recordings; and neural networks. This work will dramatically improve our understanding of language systems and test and develop a new way to model neural computation generally.

PROJECT SUMMARY / ABSTRACT Patients with Locked-In Syndrome (LIS) are cognitively intact but unable to move or communicate except through eye blinks or limited (vertical) eye movements. For many patients suffering from this predicament, self-initiated communication is no longer possible, and they are unable to start a conversation, ask a question, or draw caregiver attention. Having no other means of interacting with family or friends, these LIS patients lack privacy and any degree of independence. Recently, a Brain-Computer Interface (BCI) device was successfully implanted in a locked-in patient with late stage ALS. By modulating her brain signals, this patient was able to select letters and spell words. The breakthrough was that, for the first time, a BCI system performed well enough to enable use at home without the need for experts. This was published in the New England Journal of Medicine. Yet, the rate of spelling was only 2 characters per minute, which clearly needs to be improved. In this collaborative early-feasibility study between the University Medical Center Utrecht (The Netherlands), the Johns Hopkins University (Baltimore), and the Neuromodulation Research unit of Medtronic (Minneapolis), we will use a significantly improved version of the implant device to develop faster and more versatile BCI control techniques for home use. The overall goal of the research is 1) to prove that the basic functionality of this BCI is generalizable to other patients with LIS (AIM1), and 2) to expand current BCI capabilities through research along 3 avenues aiming to achieve A) multidimensional item selection (avenue ?Multiselect?) that would allow patients to choose among multiple menu options, B) 2D cursor control (avenue ?Navigate?) that would allow point-and-click use of a computer, similar to that of eye trackers for users with normal eye movements, and C) vocalization through a speech synthesizer (avenue ?Speak?) (AIM2), which would greatly enhance communication and quality of life. A total of nine LIS patients will participate for one year each, during and after which they may continue to use the BCI system at home. An additional 16 patients with electrode grids implanted for unrelated diagnostic purposes (epilepsy surgery) will be included for the development of advanced decoding algorithms, utilizing clinical 256-channel recording systems. The study will demonstrate the feasibility of providing LIS patients with an alternative means of communication, thereby moving the current state of the art in human neuroscience to the realm of clinical treatment for severe communication disorders.

DESCRIPTION (provided by applicant): The overall project goals are to study the cortical network dynamics of human upper limb motor control spanning two distinct spatial scales recorded with electrocorticography (ECoG), and to demonstrate that these dynamics can be estimated in real-time and used to control the JHU Applied Physics Lab Modular Prosthetic Limb (MPL) during execution of functionally useful complex action sequences. Our human subjects will be instructed to perform complete functional movements characteristic of activities of daily living. We will analyze the task-related temporal evolution in the strength and pattern of interactions among large-scale cortical networks known to be recruited in visually-guided reach-to-grasp tasks. Using multi-scale subdural ECoG with combinations of routine clinical macro-electrodes (2.3 mm diameter, 1 cm spacing) recording activity of broadly spread elements/nodes of neural networks, and inset arrays of microelectrodes (75 ?m diameter, 0.9 mm spacing) recording the activity of local sub-networks, we will test our overall hypothesis that there is a functional hierarchy between the two scales (Aim 1). More specifically, we hypothesize that large-scale network dynamics involving premotor/motor cortex reflect the evolution of sensory-motor processing demands during complex action sequences, while micro-scale population activity and network dynamics in motor cortex reflect the low-level kinematics of these tasks. We will utilize methods of estimating dynamic effective connectivity developed by our team to study interactions between these scales and test whether there exists a spatially heterogeneous and hierarchical structure within the macro-micro scale networks. The results of these analyses have wide-ranging clinical implications for both the optimal scale of functional mapping for clinical diagnostic purposes and the extent of implantations for neuroprosthetic control. We will exploit multi-scale ECoG recordings and online estimates of the dynamics of neural activation and large-scale/local network interactions to achieve control of the MPL during functionally useful tasks (Aim 2). This approach will go beyond traditional paradigms that have developed neural control over individual degrees of freedom. We will do this by embedding low-level control within an innovative framework whereby knowledge of task goals supplement direct kinematic decoding. This project will build on our team's previous successes in implementing a system for semi-autonomous ECoG control of the MPL, employing machine vision and route-planning algorithms, during complex interactions with objects requiring the coordination of multiple joints. This system will be able to leverage for the first time the rich complexity of temporally and spatially resolved network dynamics correlated with high-level goals to achieve functionally useful control of an advanced neuroprosthetic limb.

To avoid post-operative language impairments after surgery for drug-resistant epilepsy, electrical cortical stimulation mapping (ESM) is widely relied upon to localize language cortex, but ESM often elicits after- discharges (ADs), seizures, involuntary movements, and pain, which can limit mapping and impact patient safety. Furthermore, ESM is time-consuming and usually produces all-or-none results, providing limited insight into the function of stimulated sites. Finally, up to 30% of patients can have language dysfunction after resections that are guided by ESM. These limitations have motivated an alternative mapping method based on task-related power changes in high-gamma frequencies (?50-200 Hz), which are highly correlated with neuronal population firing rate changes. Although high-gamma mapping (HGM) overcomes the aforementioned limitations of ESM, its clinical validation has been limited to ESM-HGM comparisons in small case series with significant variations in technique across centers, yielding inconsistent results, with unexplained discrepancies between methods. Moreover, neither HGM nor ESM have been prospectively validated for predicting post-operative language impairments with either subdural electrocorticography (ECoG) or stereo-EEG (sEEG, increasingly a safer alternative to ECoG). The overall objective of this study is to use a small consortium of three large academic epilepsy surgery centers to demonstrate the diagnostic validity and safety of iEEG HGM for both ECoG and sEEG in a prospective series of 221 patients, using an innovative browser-based bedside HGM system to standardize methods and share data across centers. This study will test the hypothesis that HGM can accurately predict post-surgical language outcomes but requires a different framework for interpretation than that used for ESM. First, using traditional methods for interpreting HGM results, we will test the concordance between HGM and ESM, and compare their safety and feasibility. Based on previous studies, we predict that HGM will be a good, but imperfect, predictor of ESM results. However, we predict that due to ESM-related seizures and differences in mapping duration, HGM will be safer and better tolerated by patients. Second, we will test models that attempt to explain and reconcile HGM false negatives and false positives with respect to ESM, and we will test whether these models, incorporating both functional activation (HGM) and effective connectivity (measured with cortico-cortical evoked potentials, or CCEPs) can better predict ESM results. Third, we will develop and test models for prediction of post-operative language outcomes, based on the anatomical extent and volume of cortical resection with respect to HGM and ESM results, using voxel-lesion-symptom matching (VLSM). Since clinical decisions are currently based on ESM and clinicians will be blinded to HGM, we predict that the incidence of language deficits will be significantly higher when HGM+ electrodes are resected. This study will have a direct and profound impact on the clinical practice of language mapping for neurosurgical procures, while contributing valuable insights into the structure and dynamics of human language networks.

For many years brain-computer interfaces (BCI's) have been explored as a means of restoring communication to patients with Locked-In Syndrome (LIS), a devastating and often irreversible neurological condition in which cognition is intact but nearly all motor output from the brain is interrupted, effectively cutting off communication with the outside world. To date non-invasive BCI's (e.g. EEG) have had inadequate signal fidelity and spatial resolution, while invasive BCI's using microelectrode arrays in hand motor cortex have delivered cursor and multi-joint robotic control in controlled settings, but have been difficult to learn and have required frequent retraining of decoding models, due to instabilities in the microelectrode-tissue interface. High-density electrocorticographic (ECoG) recordings have been recently used by our team (JHU and University of Utrecht) and by others for real-time detection and classification of a variety of different upper limb movements and speech components. ECoG has sufficient spatial-temporal resolution and signal quality to decode the broadband high-gamma (~60-200 Hz) responses of native cortical representations for upper limb movements and speech. Speech representations are spatially distributed over several square centimeters, ideally suited for electrocorticography (ECoG), but impractical for MEA's. In a recent landmark paper in NEJM (Vansteensel et al. 2016) Dr. Ramsey's team in Utrecht demonstrated home use of a fully implantable wireless ECoG BCI by a patient with LIS, without supervision by researchers. To expand on the capabilities of this 4-channel system, our team proposes a first-in-human clinical trial to establish the safety and efficacy of an ECoG BCI with far more channels, implanted for 6 months. Based on the long-term safety and signal quality of ECoG demonstrated in neuromodulation for epilepsy (Neuropace RNS), we have an IDE for the proposed ?CortiCom System?, which uniquely combines a 128-channel HD- ECoG array (PMT Corp) with a transcutaneous pedestal connector and neural signal processor (Blackrock Microsystems). In this early feasibility trial, our team will pursue the following Aims/Milestones: 1. Demonstrate efficient and stable control of essential BCI functions (initiate BCI, call caregiver, and BCI menu navigation). CortiCom will use real-time decoding of attempted movements of different fingers, arm joints, and mouth and face muscles to control the critical BCI functions, e.g. caregiver calling (by 3 months), and menu navigation--Up/Down/Left/Right, Enter, and Back/Escape (6 commands). 2. Demonstrate efficient and stable operation of a keyword-based speech BCI. CortiCom will use low- latency detection and classification of attempted speech (keywords) to expand communication. Keyword decoding will use a hierarchical hybrid model to detect and classify keywords based on their unique spatial- temporal signatures of population activity. Keyword command vocabulary, based on communication value and ease of classification, will expand from 6 (by 3 months) to 20 or more keywords during the trial.