Closed-loop Neuromodulation Systems Research Articles

Brain Feature Extraction With an Artifact-Tolerant Multiplexed Time-Encoding Neural Frontend for True Real-Time Closed-Loop Neuromodulation.

Closed-loop neuromodulation is emerging as a more effective and targeted solution for the treatment of neurological symptoms compared to traditional open-loop stimulation. The majority of the present designs lack the ability to continuously record brain activity during electrical stimulation; hence they cannot fully monitor the treatment's effectiveness. This is due to the large stimulation artifacts that can saturate the sensitive readout circuits. To overcome this challenge, this work presents a rapid-artifact-recovery time-multiplexed neural readout frontend in combination with backend linear interpolation to reconstruct the artifact-corrupted local field potentials' (LFP) features. Our hybrid technique is an alternative approach to avoid power-hungry large-dynamic-range readout architectures or large and complex artifact template subtraction circuits. We discuss the design and measurements of a prototype implementation of the proposed readout frontend in 180-nm CMOS. It combines time multiplexing and time-domain conversion in a novel 13-bit incremental ADC, requiring only 0.0018 mm2/channel of readout area despite the large 180-nm CMOS process used, while consuming only 4.51 μW/channel. This is the smallest reported area for stimulation-voltage-compatible technologies (i.e. ≥ 65 nm). The frontend also yields a best-in-class peak total harmonic distortion of -72.6 dB @2.5-mVpp input, thanks to its implicit DAC mismatch-error shaping property. We employ the chip to measure brain LFP signals corrupted with artifacts, then perform linear interpolation and feature extraction on the measured signals and evaluate the reconstruction quality, using a set of sixteen commonly used features and three stimulation scenarios. The results show relative accuracies above 95% with respect to the situation without artifacts. This work is an ideal candidate for integration in high-channel-count true closed-loop neuromodulation systems.

Benchmarking of hardware-efficient real-time neural decoding in brain–computer interfaces

Designing processors for implantable closed-loop neuromodulation systems presents a formidable challenge owing to the constrained operational environment, which requires low latency and high energy efficacy. Previous benchmarks have provided limited insights into power consumption and latency. However, this study introduces algorithmic metrics that capture the potential and limitations of neural decoders for closed-loop intra-cortical brain–computer interfaces in the context of energy and hardware constraints. This study benchmarks common decoding methods for predicting a primate’s finger kinematics from the motor cortex and explores their suitability for low latency and high energy efficient neural decoding. The study found that ANN-based decoders provide superior decoding accuracy, requiring high latency and many operations to effectively decode neural signals. Spiking neural networks (SNNs) have emerged as a solution, bridging this gap by achieving competitive decoding performance within sub-10 ms while utilizing a fraction of computational resources. These distinctive advantages of neuromorphic SNNs make them highly suitable for the challenging closed-loop neural modulation environment. Their capacity to balance decoding accuracy and operational efficiency offers immense potential in reshaping the landscape of neural decoders, fostering greater understanding, and opening new frontiers in closed-loop intra-cortical human-machine interaction.

Functional ultrasound imaging of the human spinal cord

Utilizing the first in-human functional ultrasound imaging (fUSI) of the spinal cord, we demonstrate the integration of spinal functional responses to electrical stimulation. We record and characterize the hemodynamic responses of the spinal cord to a neuromodulatory intervention commonly used for treating pain and increasingly used for the restoration of sensorimotor and autonomic function. We found that the hemodynamic response to stimulation reflects a spatiotemporal modulation of the spinal cord circuitry not previously recognized. Our analytical capability offers a mechanism to assess blood flow changes with a new level of spatial and temporal precision invivo and demonstrates that fUSI can decode the functional state of spinal networks in a single trial, which is of fundamental importance for developing real-time closed-loop neuromodulation systems. This work is a critical step toward developing a vital technique to study spinal cord function and effects of clinical neuromodulation.

Catheter-Free Urodynamics Testing: Current Insights and Clinical Potential.

Lower urinary tract dysfunction not only interferes with the health-related quality of life of patients but may also lead to acute kidney injury and infections. To assess the bladder, urodynamic studies (UDS) have been implemented but the use of catheters leads to discomfort for the patient. Catheter-free long-term UDS would be useful and a potential solution could be ambulatory wireless devices that communicate via telemetry. Such sensors can detect pressure or volume. Numerous types of potential catheter-free sensors have been proposed for bladder monitoring. Despite substantial innovation in the manufacturing of implantable biomedical electronic systems, such sensors have remained at the laboratory stage due to a number of critical challenges. These challenges primarily concern hermeticity and biocompatibility, sensitivity and artifacts, drift, telemetry, and energy management. Having overcome these challenges, catheter-free ambulatory urodynamic monitoring could combine a synchronized intravesical pressure sensor with a volume analyzer but only the steps of cystometry and volume measurement are currently sufficiently reproducible to simulate UDS results. The measurement of volume by infrared optical sensors, in the form of abdominal patches, appears to be promising and studies are underway to market a telemetric ambulatory urodynamic monitoring system that includes an intravesical pressure sensor. There has been considerable progress in wearable and conformable electronics on many fronts, and continued collaboration between engineers and urologists could quickly overcome current challenges. In addition, to the diagnosis of UDS, such sensors could be useful in the development of a long-term closed-loop neuromodulation system. In this review, we explore the various types of catheter-free bladder sensors, inherent challenges and solutions to overcome these challenges, and the clinical potential of such long-term implantable sensors.

Inferring cognitive state underlying conflict choices in verbal Stroop task using heterogeneous input discriminative-generative decoder model

Objective. The subthalamic nucleus (STN) of the basal ganglia interacts with the medial prefrontal cortex (mPFC) and shapes a control loop, specifically when the brain receives contradictory information from either different sensory systems or conflicting information from sensory inputs and prior knowledge that developed in the brain. Experimental studies demonstrated that significant increases in theta activities (2–8 Hz) in both the STN and mPFC as well as increased phase synchronization between mPFC and STN are prominent features of conflict processing. While these neural features reflect the importance of STN-mPFC circuitry in conflict processing, a low-dimensional representation of the mPFC–STN interaction referred to as a cognitive state, that links neural activities generated by these sub-regions to behavioral signals (e.g. the response time), remains to be identified. Approach. Here, we propose a new model, namely, the heterogeneous input discriminative-generative decoder (HI-DGD) model, to infer a cognitive state underlying decision-making based on neural activities (STN and mPFC) and behavioral signals (individuals’ response time) recorded in ten Parkinson’s disease (PD) patients while they performed a Stroop task. PD patients may have conflict processing which is quantitatively (may be qualitative in some) different from healthy populations. Main results. Using extensive synthetic and experimental data, we showed that the HI-DGD model can diffuse information from neural and behavioral data simultaneously and estimate cognitive states underlying conflict and non-conflict trials significantly better than traditional methods. Additionally, the HI-DGD model identified which neural features made significant contributions to conflict and non-conflict choices. Interestingly, the estimated features match well with those reported in experimental studies. Significance. Finally, we highlight the capability of the HI-DGD model in estimating a cognitive state from a single trial of observation, which makes it appropriate to be utilized in closed-loop neuromodulation systems.

A wearable platform for closed-loop stimulation and recording of single-neuron and local field potential activity in freely moving humans

Advances in technologies that can record and stimulate deep brain activity in humans have led to impactful discoveries within the field of neuroscience and contributed to the development of novel therapies for neurological and psychiatric disorders. Further progress, however, has been hindered by device limitations in that recording of single-neuron activity during freely moving behaviors in humans has not been possible. Additionally, implantable neurostimulation devices, currently approved for human use, have limited stimulation programmability and restricted full-duplex bidirectional capability. In this study, we developed a wearable bidirectional closed-loop neuromodulation system (Neuro-stack) and used it to record single-neuron and local field potential activity during stationary and ambulatory behavior in humans. Together with a highly flexible and customizable stimulation capability, the Neuro-stack provides an opportunity to investigate the neurophysiological basis of disease, develop improved responsive neuromodulation therapies, explore brain function during naturalistic behaviors in humans and, consequently, bridge decades of neuroscientific findings across species.

Dynamic Multi-Graph Convolution based Channel-Weighted Transformer Feature Fusion Network for Epileptic Seizure Prediction.

Electroencephalogram (EEG) based seizure prediction plays an important role in the closed-loop neuromodulation system. However, most existing seizure prediction methods based on graph convolution network only focused on constructing the static graph, ignoring multi-domain dynamic changes in deep graph structure. Moreover, the existing feature fusion strategies generally concatenated coarse-grained epileptic EEG features directly, leading to the suboptimal seizure prediction performance. To address these issues, we propose a novel multi-branch dynamic multi-graph convolution based channel-weighted transformer feature fusion network (MB-dMGC-CWTFFNet) for the patient-specific seizure prediction with the superior performance. Specifically, a multi-branch (MB) feature extractor is first applied to capture the temporal, spatial and spectral representations fromthe epileptic EEG jointly. Then, we design a point-wise dynamic multi-graph convolution network (dMGCN) to dynamically learn deep graph structures, which can effectively extract high-level features from the multi-domain graph. Finally, by integrating the local and global channel-weighted strategies with the multi-head self-attention mechanism, a channel-weighted transformer feature fusion network (CWTFFNet) is adopted to efficiently fuse the multi-domain graph features. The proposed MB-dMGC-CWTFFNet is evaluated on the public CHB-MIT EEG and a private intracranial sEEG datasets, and the experimental results demonstrate that our proposed method achieves outstanding prediction performance compared with the state-of-the-art methods, indicating an effective tool for patient-specific seizure warning. Our code will be available at: https://github.com/Rockingsnow/MB-dMGC-CWTFFNet.

Automated Tuning of Closed-loop Neuromodulation Control Systems using Bayesian Optimization.

Tuning the parameters of controllers to attain the best performance is a challenging task in designing effective closed-loop neuromodulation systems. In this paper, we present a distributed architecture for automated tuning and adaptation of closed-loop neuromodulation control systems. We use this approach for the automated parameter tuning of a Proportional-Integral (PI) neuromodulation controller using Bayesian optimization. We use a biophysically-grounded mean-field model of neural populations under electrical stimulation as a simulation environment for testing and prototyping the proposed framework and characterizing its performance. Our results demonstrate the feasibility of using Bayesian optimization for performance-based automated tuning of a PI controller in closed-loop set-point neuromodulation control tasks.

Activity dependent stimulation increases synaptic efficacy in spared pathways in an anesthetized rat model of spinal cord contusion injury.

Background:Closed-loop neuromodulation systems have received increased attention in recent years as potential therapeutic approaches for treating neurological injury and disease.Objective:The purpose of this study was to assess the ability of intraspinal microstimulation (ISMS), triggered by action potentials (spikes) recorded in motor cortex, to alter synaptic efficacy in descending motor pathways in an anesthetized rat model of spinal cord injury (SCI).Methods:Experiments were carried out in adult, male, Sprague Dawley rats with a moderate contusion injury at T8. For activity-dependent stimulation (ADS) sessions, a recording microelectrode was used to detect neuronal spikes in motor cortex that triggered ISMS in the spinal cord grey matter. SCI rats were randomly assigned to one of four experimental groups differing by: a) cortical spike-ISMS stimulus delay (10 or 25 ms) and b) number of ISMS pulses (1 or 3). Four weeks after SCI, ADS sessions were conducted in three consecutive 1-hour conditioning bouts for a total of 3 hours. At the end of each conditioning bout, changes in synaptic efficacy were assessed using intracortical microstimulation (ICMS) to examine the number of spikes evoked in spinal cord neurons during 5-minute test bouts. A multichannel microelectrode recording array was used to record cortically-evoked spike activity from multiple layers of the spinal cord.Results:The results showed that ADS resulted in an increase in cortically-evoked spikes in spinal cord neurons at specific combinations of spike-ISMS delays and numbers of pulses. Efficacy in descending motor pathways was increased throughout all dorsoventral depths of the hindlimb spinal cord.Conclusions:These results show that after an SCI, ADS can increase synaptic efficacy in spared pathways between motor cortex and spinal cord. This study provides further support for the potential of ADS therapy as an effective method for enhancing descending motor control after SCI.

Neuroweaver: a platform for designing intelligent closed-loop neuromodulation systems

Neuroweaver: a platform for designing intelligent closed-loop neuromodulation systems

Ramp Rate Evaluation and Configuration for Safe and Tolerable Closed-Loop Deep Brain Stimulation.

Closed-loop deep brain stimulation is a novel form of therapy that has shown benefit in preliminary studies and may be clinically available in the near future. Initial closed-loop studies have primarily focused on responding to sensed biomarkers with adjustments to stimulation amplitude, which is often perceptible to study participants depending on the slew or "ramp" rate of the amplitude changes. These subjective responses to stimulation ramping can result in transient side effects, illustrating that ramp rate is a unique safety parameter for closed-loop neural systems. This presents a challenge to the future of closed-loop neuromodulation systems: depending on the goal of the control policy, clinicians will need to balance ramp rates to avoid side effects and keep the stimulation therapeutic by responding in time to affect neural dynamics. In this paper, we demonstrate the results of an initial investigation into methodology for finding safe and tolerable ramp rates in four people with Parkinson's disease (PD). Results suggest that optimal ramp rates were found more accurately during varying stimulation when compared to simply toggling between maximal and minimal intensity levels. Additionally, switching frequency instantaneously was tolerable at therapeutic levels of stimulation. Future work should focus on including optimization techniques to find ramp rates.

A brief demonstration of frontostriatal connectivity in OCD patients with intracranial electrodes

Closed-loop neuromodulation is presumed to be the logical evolution for improving the effectiveness of deep brain stimulation (DBS) treatment protocols (Widge et al., 2018). Identifying symptom-relevant biomarkers that provide meaningful feedback to stimulator devices is an important initial step in this direction. This report demonstrates a technique for assaying neural circuitry hypothesized to contribute to OCD and DBS treatment outcomes. We computed phase-lag connectivity between LFPs and EEGs in thirteen treatment-refractory OCD patients. Simultaneous recordings from scalp EEG and externalized DBS electrodes in the ventral capsule/ventral striatum (VC/VS) were collected at rest during the perioperative treatment stage. Connectivity strength between midfrontal EEG sensors and VC/VS electrodes correlated with baseline OCD symptoms and 12-month posttreatment OCD symptoms. Results are qualified by a relatively small sample size, and limitations regarding the conclusiveness of VS and mPFC as neural generators given some concerns about volume conduction. Nonetheless, findings are consistent with treatment-relevant tractography findings and theories that link frontostriatal hyperconnectivity to the etiopathogenesis of OCD. Findings support the continued investigation of connectivity-based assays for aiding in determination of optimal stimulation location, and are an initial step towards the identification of biomarkers that can guide closed-loop neuromodulation systems.

Closed-Loop Implantable Therapeutic Neuromodulation Systems Based on Neurochemical Monitoring.

Closed-loop or intelligent neuromodulation allows adjustable, personalized neuromodulation which usually incorporates the recording of a biomarker, followed by implementation of an algorithm which decides the timing (when?) and strength (how much?) of stimulation. Closed-loop neuromodulation has been shown to have greater benefits compared to open-loop neuromodulation, particularly for therapeutic applications such as pharmacoresistant epilepsy, movement disorders and potentially for psychological disorders such as depression or drug addiction. However, an important aspect of the technique is selection of an appropriate, preferably neural biomarker. Neurochemical sensing can provide high resolution biomarker monitoring for various neurological disorders as well as offer deeper insight into neurological mechanisms. The chemicals of interest being measured, could be ions such as potassium (K+), sodium (Na+), calcium (Ca2+), chloride (Cl−), hydrogen (H+) or neurotransmitters such as dopamine, serotonin and glutamate. This review focusses on the different building blocks necessary for a neurochemical, closed-loop neuromodulation system including biomarkers, sensors and data processing algorithms. Furthermore, it also highlights the merits and drawbacks of using this biomarker modality.

A wireless and artefact-free 128-channel neuromodulation device for closed-loop stimulation and recording in non-human primates.

Closed-loop neuromodulation systems aim to treat a variety of neurological conditions by delivering and adjusting therapeutic electrical stimulation in response to a patient's neural state, recorded in real time. Existing systems are limited by low channel counts, lack of algorithmic flexibility, and the distortion of recorded signals by large and persistent stimulation artefacts. Here, we describe an artefact-free wireless neuromodulation device that enables research applications requiring high-throughput data streaming, low-latency biosignal processing, and simultaneous sensing and stimulation. The device is a miniaturized neural interface capable of closed-loop recording and stimulation on 128 channels, with on-board processing to fully cancel stimulation artefacts. In addition, it can detect neural biomarkers and automatically adjust stimulation parameters in closed-loop mode. In a behaving non-human primate, the device enabled long-term recordings of local field potentials and the real-time cancellation of stimulation artefacts, as well as closed-loop stimulation to disrupt movement preparatory activity during a delayed-reach task. The neuromodulation device may help advance neuroscientific discovery and preclinical investigations of stimulation-based therapeutic interventions.

Detecting Bladder Biomarkers for Closed-Loop Neuromodulation: A Technological Review.

Neuromodulation was introduced for patients with poor outcomes from the existing traditional treatment approaches. It is well-established as an alternative, novel treatment option for voiding dysfunction. The current system of neuromodulation uses an open-loop system that only delivers continuous stimulation without considering the patient’s state changes. Though the conventional open-loop system has shown positive clinical results, it can cause problems such as decreased efficacy over time due to neural habituation, higher risk of tissue damage, and lower battery life. Therefore, there is a need for a closed-loop system to overcome the disadvantages of existing systems. The closed-loop neuromodulation includes a system to monitor and stimulate micturition reflex pathways from the lower urinary tract, as well as the central nervous system. In this paper, we reviewed the current technological status to measure biomarker for closed-loop neuromodulation systems for voiding dysfunction.

Neural signal processing for closed-loop neuromodulation

The purpose of this article is to provide an overview of the current status of neural signal processing techniques for closed-loop neuromodulation. First we described overall structure of closed-loop neuromodulation systems. Then, the techniques for the stimulus artifact removal were explained, and the methods for neural state monitoring and biomarker extraction were described. Finally, the current status of neuromodulation based on neural signal processing was provided in detail. Closed-loop neuromodulation system automatically adjusts stimulation parameters based on the brain response in real time. Adequate tools for signal sensing and signal processing can be used to obtain meaningful biomarkers reflecting the state of neural systems. Especially, an appropriate neural signal processing technique can optimize the details of stimulation for effective treatment of target disease. Neural signal-based biomarkers reflecting the pathophysiological statuses of patients are essential for closedloop neuromodulation, and they should be developed from an understanding of the relationship between clinical states and neural signals.

Dominant efficiency of nonregular patterns of subthalamic nucleus deep brain stimulation for Parkinson’s disease and obsessive-compulsive disorder in a data-driven computational model

Objective. Almost 30 years after the start of the modern era of deep brain stimulation (DBS), the subthalamic nucleus (STN) still constitutes a standard stimulation target for advanced Parkinson’s disease (PD), but the use of STN-DBS is also now supported by level I clinical evidence for treatment-refractory obsessive-compulsive disorder (OCD). Disruption of neural synchronization in the STN has been suggested as one of the possible mechanisms of action of standard and alternative patterns of STN-DBS at a local level. Meanwhile, recent experimental and computational modeling evidence has signified the efficiency of alternative patterns of stimulation; however, no indications exist for treatment-refractory OCD. Here, we comparatively simulate the desynchronizing effect of standard (regular at 130 Hz) versus temporally alternative (in terms of frequency, temporal variability and the existence of bursts or pauses) patterns of STN-DBS for PD and OCD, by means of a stochastic dynamical model and two microelectrode recording (MER) datasets. Approach. The stochastic model is fitted to subthalamic MERs acquired during eight surgical interventions for PD and eight surgical interventions for OCD. For each dynamical system simulated, we comparatively assess the invariant density (steady-state phase distribution) as a measure inversely related to the desynchronizing effect yielded by the applied patterns of stimulation. Main results. We demonstrate that high (130 Hz)—and low (80 Hz)—frequency irregular patterns of stimulation, and low-frequency periodic stimulation interrupted by bursts of pulses, yield in both pathologic conditions a significantly stronger desynchronizing effect compared with standard STN-DBS, and distinct alternative patterns of stimulation. In PD, values of the invariant density measure are proven to be optimal at the dorsolateral oscillatory region of the STN including sites with the optimal therapeutic window. Significance. In addition to providing novel insights into the efficiency of low-frequency nonregular patterns of STN-DBS for advanced PD and treatment-refractory OCD, this work points to a possible correlation of a model-based outcome measure with clinical effectiveness of stimulation and may have significant implications for an energy- and therapeutically-efficient configuration of a closed-loop neuromodulation system.

Synthesis of high-complexity rhythmic signals for closed-loop electrical neuromodulation

We propose an approach to synthesizing high-complexity rhythmic signals for closed-loop electrical neuromodulation using cognitive rhythm generator (CRG) networks, wherein the CRG is a hybrid oscillator comprised of (1) a bank of neuronal modes, (2) a ring device (clock), and (3) a static output nonlinearity (mapper). Networks of coupled CRGs have been previously implemented to simulate the electrical activity of biological neural networks, including in silico models of epilepsy, producing outputs of similar waveform and complexity to the biological system. This has enabled CRG network models to be used as platforms for testing seizure control strategies. Presently, we take the application one step further, envisioning therapeutic CRG networks as rhythmic signal generators creating neuromimetic signals for stimulation purposes, motivated by recent research indicating that stimulus complexity and waveform characteristics influence neuromodulation efficacy. To demonstrate this concept, an epileptiform CRG network generating spontaneous seizure-like events (SLEs) was coupled to a therapeutic CRG network, forming a closed-loop neuromodulation system. SLEs are associated with low-complexity dynamics and high phase coherence in the network. The tuned therapeutic network generated a high-complexity, multi-banded rhythmic stimulation signal with prominent theta and gamma-frequency power that suppressed SLEs and increased dynamic complexity in the epileptiform network, as measured by a relative increase in the maximum Lyapunov exponent and decrease in phase coherence. CRG-based neuromodulation outperformed both low and high-frequency periodic pulse stimulation, suggesting that neuromodulation using complex, biomimetic signals may provide an improvement over conventional electrical stimulation techniques for treating neurological disorders such as epilepsy.

A translational platform for prototyping closed-loop neuromodulation systems

While modulating neural activity through stimulation is an effective treatment for neurological diseases such as Parkinson's disease and essential tremor, an opportunity for improving neuromodulation therapy remains in automatically adjusting therapy to continuously optimize patient outcomes. Practical issues associated with achieving this include the paucity of human data related to disease states, poorly validated estimators of patient state, and unknown dynamic mappings of optimal stimulation parameters based on estimated states. To overcome these challenges, we present an investigational platform including: an implanted sensing and stimulation device to collect data and run automated closed-loop algorithms; an external tool to prototype classifier and control-policy algorithms; and real-time telemetry to update the implanted device firmware and monitor its state. The prototyping system was demonstrated in a chronic large animal model studying hippocampal dynamics. We used the platform to find biomarkers of the observed states and transfer functions of different stimulation amplitudes. Data showed that moderate levels of stimulation suppress hippocampal beta activity, while high levels of stimulation produce seizure-like after-discharge activity. The biomarker and transfer function observations were mapped into classifier and control-policy algorithms, which were downloaded to the implanted device to continuously titrate stimulation amplitude for the desired network effect. The platform is designed to be a flexible prototyping tool and could be used to develop improved mechanistic models and automated closed-loop systems for a variety of neurological disorders.