Stochastic Signal Processing Based Stimulation Artifact Cancellation in $\Delta\Sigma$ Neural Frontend

Area of Science:

• Biomedical Engineering
• Neural Engineering
• Signal Processing

Background:

• Neural recording systems are crucial for understanding brain activity.
• Electrical stimulation is often used in conjunction with neural recording but introduces artifacts.
• Existing artifact cancellation methods face challenges in efficiency and effectiveness.

Purpose of the Study:

• To develop a power-efficient neural recorder frontend with real-time electrical stimulation artifact cancellation.
• To implement an adaptive Least Mean Squares (LMS) filter in the stochastic domain for artifact compensation.
• To validate the system's performance on a prototype Application-Specific Integrated Circuit (ASIC).

Main Methods:

• Designed a low-noise analog frontend coupled with a 1st-order delta-sigma (ΔΣ) modulator.
• Developed a power-efficient stochastic signal processor to process the ΔΣ modulator output bitstream.
• Employed an adaptive LMS filter in the stochastic domain for artifact learning and compensation.

Main Results:

• Achieved a total power consumption of 6.83 μW, with the signal processor consuming only 0.51 μW.
• Demonstrated effective suppression of 200 mV peak-to-peak stimulation artifacts by approximately 33 dB over a 10 kHz bandwidth.
• Validated artifact attenuation of 25 dB for 74.3 mVpp artifacts from biphasic stimulation in in-vitro tests.

Conclusions:

• The proposed neural recorder frontend offers a state-of-the-art solution for real-time artifact cancellation.
• The stochastic domain LMS filter approach provides a power-efficient and effective method for artifact compensation.
• The system's efficacy in suppressing artifacts from both biphasic and monophasic stimulation is confirmed.