Objective: An algorithm to reliably detect and suppress the QRS complexes of the Electrocardiography (ECG) activity contaminating the Local Field Potentials (LFP) recorded via Implantable Neurostimulators (INS) and to estimate patients’ heart rhythm.

Background: The current generation of Deep Brain Stimulation (DBS) devices allow clinicians to measure patients’ LFP signals chronically and research is currently focussed on understanding how to optimally use this new information for guided programming, therapy optimization and real-time adaptation. Biomarkers have been discovered and adaptive DBS algorithms have been developed especially for Parkinson’s Disease (PD)[1]. Unfortunately, ECG activity may contaminate LFP recordings. ECG amplitudes are much larger than LFP and its spectral content overlaps with the LFP frequency bands of interest for several diseases (e.g. 3-7 Hz for dystonia and essential tremor, 13-30 Hz for PD). In [2], positions of the INS which may minimize the problem are suggested.

Method: A method previously developed for Electroencephalography (EEG) activity and published in [3] has been adapted and improved to remove QRS complexes from LFP data recorded via an investigational INS capable of both recording and stimulation at the same time. The algorithm supports QRS detection and suppression of corresponding 1 or 2 peaks (usually Q and R), optimally preserving the LFP signal. R-R intervals can be easily calculated from consecutive R peaks to provide information on the heart rhythm of the patient, and the peak width provides an estimation of the QRS length.

Results: The method has been successfully applied on 92 LFP datasets contaminated with ECG artifacts. On ECG free datasets, the method detects false positives and should not be applied. In [figure 1], the results are shown on one of the data sets. Low frequency ECG artifact is suppressed, the biomarker 7 Hz oscillation is preserved, and its prominence from the 1/f background curve is improved.

Conclusion: The algorithm allows to remove the ECG and analyze the LFP data offline, while further developments are needed for online applications. The method also allows extraction of ECG features and successive application of algorithms for arrhythmia detection and respiration rate.

References: [1] D. Piña-Fuentes et al., “Acute effects of adaptive Deep Brain Stimulation in Parkinson’s disease,” Brain Stimulat., vol. 13, no. 6, pp. 1507–1516, Nov. 2020, doi: 10.1016/j.brs.2020.07.016. [2] M. M. Sorkhabi, M. Benjaber, P. Brown, and T. Denison, “Physiological Artifacts and the Implications for Brain-Machine-Interface Design,” in 2020 IEEE International Conference on Systems, Man, and Cybernetics (SMC), Oct. 2020, pp. 1498–1498, doi: 10.1109/SMC42975.2020.9283328. [3] C. Dora and P. K. Biswal, “Correlation-based ECG Artifact Correction from Single Channel EEG using Modified Variational Mode Decomposition,” Comput. Methods Programs Biomed., vol. 183, p. 105092, Jan. 2020, doi: 10.1016/j.cmpb.2019.105092.

« Back to MDS Virtual Congress 2021

MDS Abstracts - https://www.mdsabstracts.org/abstract/denoising-local-field-potentials-from-electrocardiography-artifacts/