Implantable neural spike detection using lifting-based stationary wavelet transform.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• High data rate neural recordings face bandwidth limitations in bio-telemetry.
• Effective spike detection requires low power and minimal area for implantable devices.
• Detecting neural spikes at low signal-to-noise ratios (SNR) is challenging.

Purpose of the Study:

• To develop an efficient spike detection system for neural signals.
• To address the bandwidth bottleneck in neural data transmission.
• To meet power and area constraints for implantable neural recording systems.

Main Methods:

• Utilized a lifting-based stationary wavelet transform (SWT).
• Decomposed neural signals into 2 levels using the 'symmlet2' wavelet basis.
• Developed a hardware implementation for the proposed spike detection system.

Main Results:

• Achieved accurate spike detection down to an SNR of 2.
• The hardware implementation consumes only 6.6 μW power.
• The system occupies a small area of 0.07 mm² for 32 channels.

Conclusions:

• The lifting-based SWT spike detection system is effective for low SNR neural signals.
• The system offers a power-efficient and area-constrained solution for bio-telemetry.
• This technology can significantly improve the feasibility of long-term neural recordings.