Sridevi V. Sarma

Network dynamics of the brain and influence of the epileptic seizure onset zone

Significance In epilepsy, seizures elicit changes in the functional connectivity of the brain that shed insight into the seizures’ nature and onset zone. We investigated the brain connectivity of patients with partial epileptic seizures from continuous multiday recordings and found that (i) the connectivity defines a finite set of brain states, (ii) seizures are characterized by a consistent progression of states, and (iii) the seizure onset zone is isolated from the surrounding regions at seizure onset but becomes most connected toward seizure termination. Our results suggest that a finite-dimensional state space model may characterize the dynamics of the epileptic brain and ultimately help localize the seizure onset zone, which is currently done by clinicians through visual inspection of electrocorticographic recordings. The human brain is a dynamic networked system. Patients with partial epileptic seizures have focal regions that periodically diverge from normal brain network dynamics during seizures. We studied the evolution of brain connectivity before, during, and after seizures with graph-theoretic techniques on continuous electrocorticographic (ECoG) recordings (5.4 ± 1.7 d per patient, mean ± SD) from 12 patients with temporal, occipital, or frontal lobe partial onset seizures. Each electrode was considered a node in a graph, and edges between pairs of nodes were weighted by their coherence within a frequency band. The leading eigenvector of the connectivity matrix, which captures network structure, was tracked over time and clustered to uncover a finite set of brain network states. Across patients, we found that (i) the network connectivity is structured and defines a finite set of brain states, (ii) seizures are characterized by a consistent sequence of states, (iii) a subset of nodes is isolated from the network at seizure onset and becomes more connected with the network toward seizure termination, and (iv) the isolated nodes may identify the seizure onset zone with high specificity and sensitivity. To localize a seizure, clinicians visually inspect seizures recorded from multiple intracranial electrode contacts, a time-consuming process that may not always result in definitive localization. We show that network metrics computed from all ECoG channels capture the dynamics of the seizure onset zone as it diverges from normal overall network structure. This suggests that a state space model can be used to help localize the seizure onset zone in ECoG recordings.

Using Point Process Models to Compare Neural Spiking Activity in the Subthalamic Nucleus of Parkinson's Patients and a Healthy Primate

Placement of deep brain stimulating electrodes in the subthalamic nucleus (STN) to treat Parkinsons disease (PD) also allows the recording of single neuron spiking activity. Analyses of these unique data offer an important opportunity to better understand the pathophysiology of PD. Despite the point process nature of PD neural spiking activity, point process methods are rarely used to analyze these recordings. We develop a point process representation of PD neural spiking activity using a generalized linear model to describe long- and short-term temporal dependencies in the spiking activity of 28 STN neurons from seven PD patients and 35 neurons from one healthy primate (surrogate control) recorded, while the subjects executed a directed-hand movement task. We used the point process model to characterize each neurons bursting, oscillatory, and directional tuning properties during key periods in the task trial. Relative to the control neurons, the PD neurons showed increased bursting, increased 10-30 Hz oscillations, and increased fluctuations in directional tuning. These features, which traditional methods failed to capture accurately, were efficiently summarized in a single model in the point process analysis of each neuron. The point process framework suggests a useful approach for developing quantitative neural correlates that may be related directly to the movement and behavioral disorders characteristic of PD.

Physiology of functional and effective networks in epilepsy.

Epilepsy is a network phenomenon characterized by atypical activity during seizure both at the level of single neurons and neural populations. The etiology of epilepsy is not completely understood but a common theme among proposed mechanisms is abnormal synchronization between neuronal populations. Recent advances in novel imaging and recording technologies have enabled the inference of comprehensive maps of both the anatomical and physiological inter-relationships between brain regions. Clinical protocols established for diagnosis and treatment of epilepsy utilize both advanced neuroimaging techniques and neurophysiological data. These growing clinical datasets can be further exploited to better understand the complex connectivity patterns in the epileptic brain. In this article, we review results and insights gained from the growing body of research focused on epilepsy from a network perspective. In particular, we put an emphasis on two different notions of network connectivity: functional and effective; and studies investigating these notions in epilepsy are highlighted. We also discuss limitations and opportunities in data collection and analyses that will further our understanding of epileptic networks and the mechanisms of seizures.

Quickest detection of drug-resistant seizures: an optimal control approach.

Epilepsy affects 50 million people worldwide, and seizures in 30% of the cases remain drug resistant. This has increased interest in responsive neurostimulation, which is most effective when administered during seizure onset. We propose a novel framework for seizure onset detection that involves (i) constructing statistics from multichannel intracranial EEG (iEEG) to distinguish nonictal versus ictal states; (ii) modeling the dynamics of these statistics in each state and the state transitions; you can remove this word if there is no room. (iii) developing an optimal control-based “quickest detection” (QD) strategy to estimate the transition times from nonictal to ictal states from sequential iEEG measurements. The QD strategy minimizes a cost function of detection delay and false positive probability. The solution is a threshold that non-monotonically decreases over time and avoids responding to rare events that normally trigger false positives. We applied QD to four drug resistant epileptic patients (168 hour continuous recordings, 26–44 electrodes, 33 seizures) and achieved 100% sensitivity with low false positive rates (0.16 false positive/hour). This article is part of a Supplemental Special Issue entitled The Future of Automated Seizure Detection and Prediction.

Reinstatement of distributed cortical oscillations occurs with precise spatiotemporal dynamics during successful memory retrieval

Significance Our results represent significant contributions to understanding the neural mechanisms and spatiotemporal dynamics governing neural reinstatement in two important ways. First, by using a cued recall memory task, our paradigm offers experimental control over retrieval. We compare reinstatement during correct and incorrect retrieval, and provide evidence that retrieval recovers a gradually changing representation of temporal context. These data provide support for mental time travel hypothesized to underlie episodic memory. Second, leveraging the high temporal precision afforded by intracranial recordings, we investigate the precise timing of reinstatement and demonstrate that retrieval may reactivate cortical representations of a memory on a faster timescale than during encoding. Our data complement previous studies demonstrating faster replay of patterns associated with a prior episode. Reinstatement of neural activity is hypothesized to underlie our ability to mentally travel back in time to recover the context of a previous experience. We used intracranial recordings to directly examine the precise spatiotemporal extent of neural reinstatement as 32 participants with electrodes placed for seizure monitoring performed a paired-associates episodic verbal memory task. By cueing recall, we were able to compare reinstatement during correct and incorrect trials, and found that successful retrieval occurs with reinstatement of a gradually changing neural signal present during encoding. We examined reinstatement in individual frequency bands and individual electrodes and found that neural reinstatement was largely mediated by temporal lobe theta and high-gamma frequencies. Leveraging the high temporal precision afforded by intracranial recordings, our data demonstrate that high-gamma activity associated with reinstatement preceded theta activity during encoding, but during retrieval this difference in timing between frequency bands was absent. Our results build upon previous studies to provide direct evidence that successful retrieval involves the reinstatement of a temporal context, and that such reinstatement occurs with precise spatiotemporal dynamics.

Therapeutic mechanisms of high-frequency stimulation in Parkinson’s disease and neural restoration via loop-based reinforcement

Significance We investigated the therapeutic mechanisms of high-frequency stimulation (HFS) in Parkinson’s disease by developing a computational model of the cortico-basal ganglia-thalamo-cortical loop in normal and parkinsonian conditions under the effects of stimulation at several frequencies. We found that the stimulation injected in the loop elicits neural perturbations that travel along multiple pathways with different latencies and rendezvous in striatum (one of the basal ganglia). If the stimulation frequency is high enough, these perturbations overlap (reinforcement) and cause more regular, stimulus-locked firing patterns in striatum. Overlap is maximal at clinically relevant HFS and restores more normal activity in the remaining structures of the loop. This suggests that neural restoration and striatal reinforcement may be a therapeutic merit and mechanism of HFS, respectively. High-frequency deep brain stimulation (HFS) is clinically recognized to treat parkinsonian movement disorders, but its mechanisms remain elusive. Current hypotheses suggest that the therapeutic merit of HFS stems from increasing the regularity of the firing patterns in the basal ganglia (BG). Although this is consistent with experiments in humans and animal models of Parkinsonism, it is unclear how the pattern regularization would originate from HFS. To address this question, we built a computational model of the cortico-BG-thalamo-cortical loop in normal and parkinsonian conditions. We simulated the effects of subthalamic deep brain stimulation both proximally to the stimulation site and distally through orthodromic and antidromic mechanisms for several stimulation frequencies (20–180 Hz) and, correspondingly, we studied the evolution of the firing patterns in the loop. The model closely reproduced experimental evidence for each structure in the loop and showed that neither the proximal effects nor the distal effects individually account for the observed pattern changes, whereas the combined impact of these effects increases with the stimulation frequency and becomes significant for HFS. Perturbations evoked proximally and distally propagate along the loop, rendezvous in the striatum, and, for HFS, positively overlap (reinforcement), thus causing larger poststimulus activation and more regular patterns in striatum. Reinforcement is maximal for the clinically relevant 130-Hz stimulation and restores a more normal activity in the nuclei downstream. These results suggest that reinforcement may be pivotal to achieve pattern regularization and restore the neural activity in the nuclei downstream and may stem from frequency-selective resonant properties of the loop.

Optimal Control-Based Bayesian Detection of Clinical and Behavioral State Transitions

Accurately detecting hidden clinical or behavioral states from sequential measurements is an emerging topic in neuroscience and medicine, which may dramatically impact neural prosthetics, brain-computer interface and drug delivery. For example, early detection of an epileptic seizure from sequential electroencephalographic (EEG) measurements would allow timely administration of anticonvulsant drugs or neurostimulation, thus reducing physical impairment and risks of overtreatment. We develop a Bayesian paradigm for state transition detection that combines optimal control and Markov processes. We define a hidden Markov model of the state evolution and develop a detection policy that minimizes a loss function of both probability of false positives and accuracy (i.e., lag between estimated and actual transition time). Our strategy automatically adapts to each newly acquired measurement based on the state evolution model and the relative loss for false positives and accuracy, thus resulting in a time varying threshold policy. The paradigm was used in two applications: 1) detection of movement onset (behavioral state) from subthalamic single unit recordings in Parkinsons disease patients performing a motor task; 2) early detection of an approaching seizure (clinical state) from multichannel intracranial EEG recordings in rodents treated with pentylenetetrazol chemoconvulsant. Our paradigm performs significantly better than chance and improves over widely used detection algorithms.

The effects of cues on neurons in the basal ganglia in Parkinson's disease

Visual cues open a unique window to the understanding of Parkinsons disease (PD). These cues can temporarily but dramatically improve PD motor symptoms. Although details are unclear, cues are believed to suppress pathological basal ganglia (BG) activity through activation of corticostriatal pathways. In this study, we investigated human BG neurophysiology under different cued conditions. We evaluated bursting, 10–30 Hz oscillations (OSCs), and directional tuning (DT) dynamics in the subthalamic nucleus (STN) activity while seven patients executed a two-step motor task. In the first step (predicted +cue), the patient moved to a target when prompted by a visual go cue that appeared 100% of the time. Here, the timing of the cue is predictable and the cue serves an external trigger to execute a motor plan. In the second step, the cue appeared randomly 50% of the time, and the patient had to move to the same target as in the first step. When it appeared (unpredicted +cue), the motor plan was to be triggered by the cue, but its timing was not predictable. When the cue failed to appear (unpredicted −cue), the motor plan was triggered by the absence of the visual cue. We found that during predicted +cue and unpredicted −cue trials, OSCs significantly decreased and DT significantly increased above baseline, though these modulations occurred an average of 640 ms later in unpredicted −cue trials. Movement and reaction times were comparable in these trials. During unpredicted +cue trials, OSCs, and DT failed to modulate though bursting significantly decreased after movement. Correspondingly, movement performance deteriorated. These findings suggest that during motor planning either a predictably timed external cue or an internally generated cue (generated by the absence of a cue) trigger the execution of a motor plan in premotor cortex, whose increased activation then suppresses pathological activity in STN through direct pathways, leading to motor facilitation in PD.

Non-stationary discharge patterns in motor cortex under subthalamic nucleus deep brain stimulation

Deep brain stimulation (DBS) of the subthalamic nucleus (STN) directly modulates the basal ganglia (BG), but how such stimulation impacts the cortex upstream is largely unknown. There is evidence of cortical activation in 6-hydroxydopamine (OHDA)-lesioned rodents and facilitation of motor evoked potentials in Parkinsons disease (PD) patients, but the impact of the DBS settings on the cortical activity in normal vs. Parkinsonian conditions is still debated. We use point process models to analyze non-stationary activation patterns and inter-neuronal dependencies in the motor and sensory cortices of two non-human primates during STN DBS. These features are enhanced after treatment with 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), which causes a consistent PD-like motor impairment, while high-frequency (HF) DBS (i.e., ≥100 Hz) strongly reduces the short-term patterns (period: 3–7 ms) both before and after MPTP treatment, and elicits a short-latency post-stimulus activation. Low-frequency DBS (i.e., ≤50 Hz), instead, has negligible effects on the non-stationary features. Finally, by using tools from the information theory [i.e., receiver operating characteristic (ROC) curve and information rate (IR)], we show that the predictive power of these models is dependent on the DBS settings, i.e., the probability of spiking of the cortical neurons (which is captured by the point process models) is significantly conditioned on the timely delivery of the DBS input. This dependency increases with the DBS frequency and is significantly larger for high- vs. low-frequency DBS. Overall, the selective suppression of non-stationary features and the increased modulation of the spike probability suggest that HF STN DBS enhances the neuronal activation in motor and sensory cortices, presumably because of reinforcement mechanisms, which perhaps involve the overlap between feedback antidromic and feed-forward orthodromic responses along the BG-thalamo-cortical loop.

Closed-loop control of a fragile network: application to seizure-like dynamics of an epilepsy model

It has recently been proposed that the epileptic cortex is fragile in the sense that seizures manifest through small perturbations in the synaptic connections that render the entire cortical network unstable. Closed-loop therapy could therefore entail detecting when the network goes unstable, and then stimulating with an exogenous current to stabilize the network. In this study, a non-linear stochastic model of a neuronal network was used to simulate both seizure and non-seizure activity. In particular, synaptic weights between neurons were chosen such that the networks fixed point is stable during non-seizure periods, and a subset of these connections (the most fragile) were perturbed to make the same fixed point unstable to model seizure events; and, the model randomly transitions between these two modes. The goal of this study was to measure spike train observations from this epileptic network and then apply a feedback controller that (i) detects when the network goes unstable, and then (ii) applies a state-feedback gain control input to the network to stabilize it. The stability detector is based on a 2-state (stable, unstable) hidden Markov model (HMM) of the network, and detects the transition from the stable mode to the unstable mode from using the firing rate of the most fragile node in the network (which is the output of the HMM). When the unstable mode is detected, a state-feedback gain is applied to generate a control input to the fragile node bringing the network back to the stable mode. Finally, when the network is detected as stable again, the feedback control input is switched off. High performance was achieved for the stability detector, and feedback control suppressed seizures within 2 s after onset.