Nenad B. Popović | Searcharxiv

arXiv2024

Data Augmentation for Generating Synthetic Electrogastrogram Time Series

To address an emerging need for large number of diverse datasets for rigor evaluation of signal processing techniques, we developed and evaluated a new method for generating synthetic electrogastrogram time series. We used electrogastrography (EGG) data from an open database to set model parameters and statistical tests to evaluate synthesized data. Additionally, we illustrated method customization for generating artificial EGG time series alterations caused by the simulator sickness. Proposed data augmentation method generates synthetic EGG data with specified duration, sampling frequency, recording state (postprandial or fasting state), overall noise and breathing artifact injection, and pauses in the gastric rhythm (arrhythmia occurrence) with statistically significant difference between postprandial and fasting states in > 70% cases while not accounting for individual differences. Features obtained from the synthetic EGG signal resembling simulator sickness occurrence displayed expected trends. The code for generation of synthetic EGG time series is not only freely available and can be further customized to assess signal processing algorithms but also may be used to increase data diversity for training artificial intelligence (AI) algorithms. The proposed approach is customized for EGG data synthesis but can be easily utilized for other biosignals with similar nature such as electroencephalogram.

eess.SP↗

arXiv2020

Combined approach for automatic and robust calculation of dominant frequency of electrogastrogram

We present a novel method for automatic and robust detection of dominant frequency (DF) in the electrogastrogram (EGG). Our new approach combines Fast Fourier Transform (FFT), Welch's method for spectral density estimation, and autocorrelation. The proposed combined method as well as other separate procedures were tested on a freely available dataset consisted of EGG recordings in 20 healthy individuals. DF was calculated in relation (1) to the fasting and postprandial states, (2) to the three recording locations, and (3) to the subjects' body mass index. For the estimation of algorithms performance in the presence of noise, we created a synthetic dataset by adding white Gaussian noise to the artifact-free EGG waveform in one subject. The individual algorithms and novel combined approach were evaluated in relation to the signal-to-noise ratio (SNR) in range from -40 dB to 20 dB. Our results showed that the novel combined method significantly outperformed the commonly used approach for DF calculation - FFT in noise presence when compared to the benchmark data being was manually corrected by an expert. The novel method outperformed autocorrelation and Welch's method in accuracy. Additionally, we presented a method for optimal window width selection when using Welch's spectrogram that showed that for DF detection, window length of N/4 (300 s), where N is the length of EGG waveform in samples, performed the best when compared to the benchmark data. The combined approach proved efficient for automatic and robust calculation of dominant frequency on openly available EGG dataset recorded in healthy individuals and is promising approach for DF detection.

eess.SP↗