A feasibility study on AI-controlled closed-loop electrical stimulation implants.

Miniaturized electrical stimulation (ES) implants show great promise in practice, but their real-time control by means of biophysical mechanistic algorithms is not feasible due to computational complexity. Here, we study the feasibility of more computationally efficient machine learning methods to control ES implants. For this, we estimate the normalized twitch force of the stimulated extensor digitorum longus muscle on n = 11 Wistar rats with intra- and cross-subject calibration. After 2000 training stimulations, we reach a mean absolute error of 0.03 in an intra-subject setting and 0.2 in a cross-subject setting with a random forest regressor. To the best of our knowledge, this work is the first experiment showing the feasibility of AI to simulate complex ES mechanistic models. However, the results of cross-subject training motivate more research on error reduction methods for this setting.

Pulse Shaping Strategies for Electroceuticals: A Comprehensive Survey of the Use of Interphase Gaps in Miniature Stimulation Systems.

OBJECTIVE: Interphase gaps (IPGs) are among the most commonly suggested pulse shape variations to try to enhance neural stimulation efficiency by reducing the action potential (AP) suppressing effect of an early anodic hyperpolarization. The majority of published literature on the effect of IPGs is based on investigations of monopolar stimulation configurations. However, many contemporary neuromodulation applications including the emerging field of electroceutical devices operate in a bipolar electrode configuration. METHODS: We investigated the effect of IPGs and asymmetric biphasic current controlled pulses with reduced anodic amplitude on neural activation in both principal electrode configurations in a rodent in-vivo nerve muscle preparation. RESULTS: In the monopolar electrode configuration, our findings of 10.9 ± 1.5% decreased stimulation amplitude with 200 µs IPGs in biphasic pulses of 40 µs phase width are in agreement with published literature in this configuration. Surprisingly, using the bipolar configuration, opposite effects of IPGs were observed and neural activation required up to 18.6 ± 3.1% (phase width 100 µs, IPG = 1000 µs) higher amplitudes. Electroneurogram recordings of the stimulated nerve revealed temporal differences in AP generation between the monopolar and bipolar configuration. In the bipolar configuration excitation first occurred in response to the middle field transition of biphasic pulses. CONCLUSION: This is the first study to report consistently increased amplitude requirements with IPGs in bipolar stimulation configurations. SIGNIFICANCE: Our findings must be taken into consideration when designing stimulation waveforms for neuromodulation devices that operate in a bipolar mode to avoid increased amplitude requirements that result in increased energy consumption.

Reply to "Energy Optimal Stimulation Waveforms, Or Not: Comments on 'An Investigation of Neural Stimulation Efficiency With Gaussian Waveforms'".

We thank Professors Grill and Wongsarnpigoon for their detailed responses prompted by our recent publication [1] in this journal. We welcome the opportunity we have had for discussion and thank the Editor for his invitation to make this further contribution. We made our experiments partly in the light of the statements in the abstract of their relevant paper [2] that energy optimal genetic algorithm (GA) waveforms "resembled truncated Gaussian curves" and that "if used in implantable neural stimulators, GA-optimized waveforms could prolong battery life, thereby reducing the frequency of recharge intervals, the volume of implanted pulse generators, and the costs and risks of battery-replacement surgeries." The authors discussed the practical consequences of both a 5% increase and a 60% increase in energy efficiency in the final statement of the paper and acknowledged that to include the cost of generating the GA waveforms might change the predicted optimal shape. Although mentioned only once, a claim of up to 60% relative energy savings is of great interest to those who design and use neural stimulators. Based on our in-vivo investigation of neural stimulation efficiency with biphasic Gaussian waveforms, we reported an upper potential of 17% energy savings over rectangular biphasic pulses. Furthermore, we added the consideration of instantaneous power requirement in our paper to further inform this discussion. This is a crucial parameter, especially to set the practical performance of the sharp-peaked GA waveforms into perspective, given that they deliver the majority of the cathodic charge in a time much shorter than the nominal phase duration.

The Effect of Sub-Threshold Pre-Pulses on Neural Activation Depends on Electrode Configuration.

OBJECTIVE: Published research on nerve stimulation with sub-threshold conditioning pre-pulses is contradictory. Like most early research on electrical stimulation (ES), the pioneer work on the use of pre-pulses was modelled and measured only for monopolar electrodes. However, many contemporary ES applications, including miniaturized neuromodulation implants, known as electroceuticals, operate in bipolar mode. METHODS: We compared depolarizing (DPPs) and hyperpolarizing (HPPs) pre-pulses on neural excitability in rat nerve with monopolar and bipolar electrodes. The rat common peroneal nerve was stimulated with biphasic stimuli with and without ramp and square DPPs or HPPs of 1, 5 and 10 ms duration and 10%-20% of the amplitude of the following pulse. RESULTS: The effects were opposite for the monopolar and bipolar configurations. With monopolar electrodes DPPs increased the amplitude required to activate 50% of the motoneuron pool (between 0.7% and 10.3%) and HPPs decreased the threshold (between 1.7% and 4.7%). With bipolar electrodes both pre-pulse types had the opposite effect: DPPs decreased thresholds (between 1.8% and 5.5%) whereas HPPs increased thresholds (between 0.5% and 4.1%). Electroneurograms from the stimulated nerve revealed spatial and temporal differences in action potential generation for monopolar and bipolar electrodes. In bipolar biphasic stimulation, excitation first occurred at the return electrode as a response to the transition between the cathodic and anodic phase. CONCLUSION: These data help to resolve the contradictions in the published data over two decades. SIGNIFICANCE: They also show that fundamental research carried out in monopolar configuration is not directly applicable to contemporary bipolar ES applications.

An Investigation of Neural Stimulation Efficiency With Gaussian Waveforms.

OBJECTIVE: Previous computational studies predict that Gaussian shaped waveforms use the least energy to activate nerves. The primary goal of this study was to examine the claimed potential of up to 60% energy savings with these waveforms over a range of phase widths (50- [Formula: see text]) in an animal model. METHODS: The common peroneal nerve of anaesthetized rats was stimulated via monopolar and bipolar electrodes with single stimuli. The isometric peak twitch force of the extensor digitorum longus muscle was recorded to indicate the extent of neural activation. The energy consumption, charge injection and maximum instantaneous power values required to reach 50% neural activation were compared between Gaussian pulses and standard rectangular stimuli. RESULTS: Energy savings in the 50- [Formula: see text] range of phase widths did not exceed 17% and were accompanied by significant increases in maximum instantaneous power of 110-200%. Charge efficiency was found to be increased over the whole range of tested phase widths with Gaussian compared to rectangular pulses and reached up to 55% at 1ms phase width. CONCLUSION: These findings challenge the claims of up to 60% energy savings with Gaussian like stimulation waveforms. The moderate energy savings achieved with the novel waveform are accompanied with considerable increases in maximal instantaneous power. Larger power sources would therefore be required, and this opposes the trend for implant miniaturization. SIGNIFICANCE: This is the first study to comprehensively investigate stimulation efficiency of Gaussian waveforms. It sheds new light on the practical potential of such stimulation waveforms.