Baseline Correction of Intraoperative Electromyography using Discrete Wavelet Transform

Abstract

In intraoperative analysis of electromygraphic signals (EMG) for monitoring purposes, baseline artefacts frequently pose considerable problems. Since artefact sources in the operating room can only be reduced to a limited degree, signal-processing methods are needed to correct the registered data online without major changes to the relevant data itself. We describe a method for baseline correction based on "discrete wavelet transform" (DWT) and evaluate its performance compared to commonly used digital filters. EMG data from 10 patients who underwent removal of acoustic neuromas were processed. Effectiveness, preservation of relevant EMG patterns and processing speed of a DWT based correction method was assessed and compared to a range of commonly used Butterworth, Resistor-Capacitor and Gaussian filters. Butterworth and DWT filters showed better performance regarding artefact correction and pattern preservation compared to Resistor-Capacitor and Gaussian filters. Assuming equal weighting of both characteristics, DWT outperformed the other methods: While Butterworth, Resistor-Capacitor and Gaussian provided good pattern preservation, the effectiveness was low and vice versa, while DWT baseline correction at level 6 performed well in both characteristics. The DWT method allows reliable and efficient intraoperative baseline correction in real-time. It is superior to commonly used methods and may be crucial for intraoperative analysis of EMG data, for example for intraoperative assessment of facial nerve function.