Concurrent stimulation and sensing in bi-directional brain interfaces: a multi-site translational experience

Objective. To provide a design analysis and guidance framework for the implementation of concurrent stimulation and sensing during adaptive deep brain stimulation (aDBS) with particular emphasis on artifact mitigations. Approach. We defined a general architecture of feedback-enabled devices, identified key components in the signal chain which might result in unwanted artifacts and proposed methods that might ultimately enable improved aDBS therapies. We gathered data from research subjects chronically-implanted with an investigational aDBS system, Summit RC + S, to characterize and explore artifact mitigations arising from concurrent stimulation and sensing. We then used a prototype investigational implantable device, DyNeuMo, and a bench-setup that accounts for tissue–electrode properties, to confirm our observations and verify mitigations. The strategies to reduce transient stimulation artifacts and improve performance during aDBS were confirmed in a chronic implant using updated configuration settings. Main results. We derived and validated a ‘checklist’ of configuration settings to improve system performance and areas for future device improvement. Key considerations for the configuration include (a) active instead of passive recharge, (b) sense-channel blanking in the amplifier, (c) high-pass filter settings, (d) tissue–electrode impedance mismatch management, (e) time-frequency trade-offs in the classifier, (f) algorithm blanking and transition rate limits. Without proper channel configuration, the aDBS algorithm was susceptible to limit-cycles of oscillating stimulation independent of physiological state. By applying the checklist, we could optimize each block’s performance characteristics within the overall system. With system-level optimization, a ‘fast’ aDBS prototype algorithm was demonstrated to be feasible without reentrant loops, and with noise performance suitable for subcortical brain circuits. Significance. We present a framework to study sources and propose mitigations of artifacts in devices that provide chronic aDBS. This work highlights the trade-offs in performance as novel sensing devices translate to the clinic. Finding the appropriate balance of constraints is imperative for successful translation of aDBS therapies.

1. Background 1.1. Susceptibility, how artifacts can limit performance In an adaptive or closed-loop DBS system, stimulation is adjusted based on control feedback signals from neural biomarkers [1] or behavioral sensors [2]. In the scheme of aDBS based on concurrent neural stimulation and sensing, the correct modulation of stimulation is susceptible to artifactual coupling in the signal chain (main manuscript Figure 2 and Figure 3). Three main types of signal artifacts can degrade the neural (biomarker) signal in a chronic stimulation and sensing device [3]: a) concurrent stimulation [4][5][6][7], b) electrocardiogram (ECG or EKG) artifacts [8,9] and c) movement artifacts [10]. Any combination of these artifacts degrades or precludes sensing fidelity and detection of the biomarkers metric for detection algorithms. For example, during beta-based 'fast' adaptive stimulation in PD, algorithm performance can be challenged due to close proximity of the stimulation and sensing electrodes (subcortical DBS lead) and the relatively low signal levels of subthalamic beta bursts (1 to 20 microvolts). Most critical can be the coupling of transient step responses through the sensing channel during stimulation amplitude ramping (increments/decrements), with ranges of transient artifacts from 50 to 500 microvolts or higher (much higher than the field potential signal). Design choices within each block of the system can either help mitigate or exacerbate these artifacts.

General design considerations of the signal chain
The canonical signal chain provides a model for assessing sensitivity to artifacts. A key consideration of the management of sense-stimulation interactions within the tissue-electrode interface, which is the key point of coupling stimulation into the sensing chain. The electrodes are placed into neural substrates with variable impedance characteristics (Supplementary Material Figure S1(a)). The relationship of stimulation dipole fields to the differential sensing dipole is critical, and the tissue differences and electrode characteristics, resulting in impedance mismatch, can allow for the stimulation dipole to contaminate the measurement ( Figure S1(b)). Choices made in the electronics interfaces --both the stimulation waveforms, as well as sensing interface --influence the degree to which artifacts impact sensing fidelity ( Figure S1C).
Additional constraints beyond sensing algorithms further inform the system design. In designing a closed-loop adaptive stimulation device safety and power management are key aspects and should comply with regulatory requirements [6]. One key safety element is the stimulation source which carries the transfer of chemical energy from battery to the electrode-tissue interface in the form of a stimulation waveform [11]. To ensure no direct current (DC) charge is transferred to the tissue, which could harm electrodes and tissue, blocking DC capacitors are typically integrated in the input/output circuitry of the device ( Figure S1(c)). In a multichannel stimulation and sensing intracranial device, this safety requirement applies also to the sense channel, thus a blocking DC capacitor may also be part of the sense engine, which may also contain additional capacitor-resistor circuits serving as high pass filters. Interactions between stimulation waveform, mismatch of electrode-tissue impedances and filter characteristics of the amplifier chain will result in a transient (step) response ( Figure 3(b)). These artifactual transient responses during variable stimulation and sensing will limit biomarker detection fidelity of a closed-loop system.

Benchtop Model and Setup
To facilitate a standardized, reproducible, and distributable lab testing environment, a benchtop electrode-tissue interface model was developed. The model was designed to optionally include any combination of the output impedance of the stimulator device, the return path impedance to the stimulator device case, and the electrode-tissue impedance for bipolar electrodes. All impedances were modeled as RC elements (an architecture published by [12][13][14]. The circuit was designed and simulated in LTSpice XVII (Analog Devices, Wilmington MA, USA). The modeled circuit is presented in Supplementary Materials Figure S5. In the initial stages of development, three use-cases were defined to the model: 1) to evaluate the performance of a closed-loop aDBS algorithm when physiologically relevant biosignals were provided, 2) to observe the response of closed-loop aDBS algorithms to physiologically relevant biosignals in the presence of an impedance mismatch, and 3) to "play back" recorded biosignals (where the electrode-tissue interface impedance effects were already present on the signal).
The model was designed to integrate as an attachment to the NeuroDAC [15], which handled the digital to analog conversion required to playback biosignals to an appropriate reproduction accuracy. Rather than simply providing one set of "balanced" and "mismatched" impedances respectively, the model was designed to facilitate on-the-fly switching of electrode and case impedances. A set of 7 impedance options for 8 circuit elements is provided, giving a total number of impedance combinations of 56 per bipolar electrode pair that can be rapidly selected without the need for soldering/desoldering. Input and output connections were selected to be stimulator agnostic. That is, the stimulator connections, and impedance outputs can connect to a great number of stimulating / sensing electrodes using simple BNC adapters. The benchtop PCB design is presented in Supplementary Materials Figure S5 (c).     Figure S1. Canonical signal chain providing a model for assessing sensitivity to artifacts. A) Electrodes are placed into neural substrates with variable impedance characteristics. This sketch represents a subcortical lead in the GP with contact C2 at the intersection of the external and internal structures GPe and GPi, B) relationship of stimulation dipole fields to the differential sensing dipole is critical, and the tissue differences and electrode characteristics, resulting in impedance mismatch, can allow for the stimulation dipole to contaminate the measurement (common mode artifact). C) The differential (local field potential) and common mode signal (in this example the stimulation artifact) interface with the sensing circuitry via the tissue-electrode impedance. Choices in the electronic interfaces -both the stimulation waveforms, as well as sensing interface -influence the degree to which artifacts impact sensing fidelity and how each independent block can be tuned for artifact mitigation.       Figure S6 and Table S3 (e.g., R = 82kΩ in parallel with C = 680nF-2.2uF and in series with R =1.5kΩ). The Stimulator case return impedances has the same conifugration with only difference in the series resistor R set to 100Ω. The stimulator current source is modeled by I1, and the NeuroDAC biosignal source is modeled by V1. Details of the design specification files are found on https://github.com/openmind-consortium/NeuroDACLoadPCB. Figure S6. RC-R tissue-electrode interface star network setup for benchtop testing (see Table S3 for values).