NeuroDAC: an open-source arbitrary biosignal waveform generator.

Objective.Researchers are developing biomedical devices with embedded closed-loop algorithms for providing advanced adaptive therapies. As these devices become more capable and algorithms become more complex, tasked with integrating and interpreting multi-channel, multi-modal electrophysiological signals, there is a need for flexible bench-top testing and prototyping. We present a methodology for leveraging off-the-shelf audio equipment to construct a biosignal waveform generator capable of streaming pre-recorded biosignals from a host computer. By re-playing known, well-characterized, but physiologically relevant real-world biosignals into a device under test, researchers can evaluate their systems without the need for expensivein vivoexperiments.Approach.An open-source design based on the proposed methodology is described and validated, the NeuroDAC. NeuroDAC allows for 8 independent channels of biosignal playback using a simple, custom designed attenuation and buffering circuit. Applications can communicate with the device over a USB interface using standard audio drivers. On-board analog amplitude adjustment is used to maximize the dynamic range for a given signal and can be independently tuned for each channel.Main results.Low noise component selection yields a no-signal noise floor of just 5.35 ± 0.063. NeuroDAC's frequency response is characterized with a high pass -3 dB rolloff at 0.57 Hz, and is capable of accurately reproducing a wide assortment of biosignals ranging from EMG, EEG, and ECG to extracellularly recorded neural activity. We also present an application example using the device to test embedded algorithms on a closed-loop neural modulation device, the Medtronic RC+S.Significance.By making the design of NeuroDAC open-source we aim to present an accessible tool for rapidly prototyping new biomedical devices and algorithms than can be easily modified based on individual testing needs.ClinicalTrials.gov Identifiers: NCT04281134, NCT03437928, NCT03582891.

Decoding hindlimb kinematics from primate motor cortex using long short-term memory recurrent neural networks.

Recent machine learning techniques have become a powerful tool in a variety of tasks, including neural decoding. Deep neural networks, particularly recurrent models, leverage the temporal evolution of neural ensemble activity to decode complex movement and sensory signals. Using single-unit recordings from microelectrode arrays implanted in the leg area of primary motor cortex in non-human primates, we decode the positions and angles of hindlimb joints during a locomotion task using a long short-term memory (LSTM) network. The LSTM decoder improved decoding over traditional filtering methods, such as Wiener and Kalman filters. However, dramatic improvements over other machine learning (e.g. XGBoost) and latent state-space methods were not observed.

Active microelectronic neurosensor arrays for implantable brain communication interfaces.

We have built a wireless implantable microelectronic device for transmitting cortical signals transcutaneously. The device is aimed at interfacing a cortical microelectrode array to an external computer for neural control applications. Our implantable microsystem enables 16-channel broadband neural recording in a nonhuman primate brain by converting these signals to a digital stream of infrared light pulses for transmission through the skin. The implantable unit employs a flexible polymer substrate onto which we have integrated ultra-low power amplification with analog multiplexing, an analog-to-digital converter, a low power digital controller chip, and infrared telemetry. The scalable 16-channel microsystem can employ any of several modalities of power supply, including radio frequency by induction, or infrared light via photovoltaic conversion. As of the time of this report, the implant has been tested as a subchronic unit in nonhuman primates ( approximately 1 month), yielding robust spike and broadband neural data on all available channels.