A Comparison of Delay-and-Add and Maximum Likelihood Estimation for Velocity-Selective Recording Using...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Neural Prosthetics

Background:

• Extracting peripheral nervous system information is crucial for neural prostheses.
• Linear electrode arrays offer temporal and spatial selectivity for neural recordings.
• Velocity selective recording aids in classifying nerve fiber types.

Purpose of the Study:

• Compare the delay-and-add and maximum likelihood estimation algorithms for velocity selective recording.
• Evaluate algorithm performance in electroneurography using in-vivo recordings.
• Assess the impact of channel count on algorithm accuracy and efficiency.

Main Methods:

• In-vivo recordings of electrically evoked compound action potentials from a pig's ulnar nerve.
• Application of delay-and-add and maximum likelihood estimation algorithms.
• Performance assessment using velocity quality factor (Q-factor), computational time, and channel number.

Main Results:

• Algorithm performance is significantly influenced by the number of recording channels.
• Accuracy increased from 77% (2 channels) to 98% (11 channels).
• Both algorithms demonstrated comparable accuracy and Q-factor, with delay-and-add showing a slight Q-factor advantage.

Conclusions:

• Both delay-and-add and maximum likelihood estimation are effective for velocity selective recording in electroneurography.
• Increasing the number of channels in recording arrays substantially improves classification accuracy.
• These findings advance the development of closed-loop neural prostheses by enhancing nerve signal analysis.