ENI
Embedded Neural Interface Framework
1,024-channel neural signal acquisition, real-time spike detection, and AI-powered neural decoding for brain-computer interfaces and closed-loop neurostimulation.
Signal Acquisition Modes
ENI supports four neural signal types with hardware-optimized acquisition pipelines.
ENI Processing Pipeline
Acquisition
1,024-channel simultaneous neural signal acquisition at up to 100 kHz per channel. 24-bit ADC resolution with <2μV RMS noise floor.
Filtering
Hardware-accelerated bandpass filtering (0.1 Hz – 10 kHz), notch filtering (50/60 Hz), and common-average referencing.
Spike Detection
Real-time spike detection using threshold crossing and template matching. Processes 1,024 channels in <500μs on Cortex-M55.
Feature Extraction
Spike waveform features, LFP power bands (delta/theta/alpha/beta/gamma), and coherence measures extracted per channel.
Decoding
EAI-powered neural decoder: population vector decoding, Kalman filter, or deep learning decoder (LSTM/Transformer) for BCI applications.
Output
Decoded neural commands delivered via EIPC to EoS applications with <10ms end-to-end latency from spike to command.
Use Cases
Brain-Computer Interface
Motor cortex decoding for prosthetic limb control. 95%+ decode accuracy for 8-direction cursor control.
Closed-Loop Neurostimulation
Real-time feedback loop: detect neural biomarkers → trigger stimulation → measure response. For epilepsy, Parkinson's, depression.
Neural Prosthetics
Cochlear implant signal processing, retinal prosthetics, and somatosensory feedback for prosthetic hands.
Research Platforms
High-density Utah array and Michigan probe support. Compatible with Open Ephys, Blackrock, and Intan hardware.