AUTHOR=Ramadan Ahmed , König Seth D. , Zhang Mingming , Ross Erika K. , Herman Alexander , Netoff Theoden I. , Darrow David P. TITLE=Methods and system for recording human physiological signals from implantable leads during spinal cord stimulation JOURNAL=Frontiers in Pain Research VOLUME=4 YEAR=2023 URL=https://www.frontiersin.org/journals/pain-research/articles/10.3389/fpain.2023.1072786 DOI=10.3389/fpain.2023.1072786 ISSN=2673-561X ABSTRACT=Objectives

This article presents a method–including hardware configuration, sampling rate, filtering settings, and other data analysis techniques–to measure evoked compound action potentials (ECAPs) during spinal cord stimulation (SCS) in humans with externalized percutaneous electrodes. The goal is to provide a robust and standardized protocol for measuring ECAPs on the non-stimulation contacts and to demonstrate how measured signals depend on hardware and processing decisions.

Methods

Two participants were implanted with percutaneous leads for the treatment of chronic pain with externalized leads during a trial period for stimulation and recording. The leads were connected to a Neuralynx ATLAS system allowing us to simultaneously stimulate and record through selected electrodes. We examined different hardware settings, such as online filters and sampling rate, as well as processing techniques, such as stimulation artifact removal and offline filters, and measured the effects on the ECAPs metrics: the first negative peak (N1) time and peak-valley amplitude.

Results

For accurate measurements of ECAPs, the hardware sampling rate should be least at 8 kHz and should use a high pass filter with a low cutoff frequency, such as 0.1 Hz, to eliminate baseline drift and saturation (railing). Stimulation artifact removal can use a double exponential or a second-order polynomial. The polynomial fit is 6.4 times faster on average in computation time than the double exponential, while the resulting ECAPs’ N1 time and peak-valley amplitude are similar between the two. If the baseline raw measurement drifts with stimulation, a median filter with a 100-ms window or a high pass filter with an 80-Hz cutoff frequency preserves the ECAPs.

Conclusions

This work is the first comprehensive analysis of hardware and processing variations on the observed ECAPs from SCS leads. It sets recommendations to properly record and process ECAPs from the non-stimulation contacts on the implantable leads.