2021/05/13 by Marco Antonio Pinto-Orellana, Pinto-Orellana, Marco A., Habib Sherkat +5
Engineering · Medicine · Neuroscience · #EEG and Brain-Computer Interfaces #FOS: Computer and information sciences #FOS: Electrical engineering #FOS: Physical sciences #Medical Physics (physics.med-ph) #Methodology (stat.ME) #Non-Invasive Vital Sign Monitoring #Optical Imaging and Spectroscopy Techniques #Signal Processing (eess.SP) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2105.10406
openalex publication_date 2021/05/13 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
This paper introduces a new time-frequency representation method for\nbiomedical signals: the dyadic aggregated autoregressive (DASAR) model.\nSignals, such as electroencephalograms (EEGs) and functional near-infrared\nspectroscopy (fNIRS), exhibit physiological information through time-evolving\nspectrum components at specific frequency intervals: 0-50 Hz (EEG) or 0-150 mHz\n(fNIRS). Spectrotemporal features in signals are conventionally estimated using\nshort-time Fourier transform (STFT) and wavelet transform (WT). However, both\nmethods may not offer the most robust or compact representation despite their\nwidespread use in biomedical contexts. The presented method, DASAR, improves\nprecise frequency identification and tracking of interpretable frequency\ncomponents with a parsimonious set of parameters. DASAR achieves these\ncharacteristics by assuming that the biomedical time-varying spectrum comprises\nseveral independent stochastic oscillators with (piecewise) time-varying\nfrequencies. Local stationarity can be assumed within dyadic subdivisions of\nthe recordings, while the stochastic oscillators can be modeled with an\naggregation of second-order autoregressive models (ASAR). DASAR can provide a\nmore accurate representation of the (highly contrasted) EEG and fNIRS frequency\nranges by increasing the estimation accuracy in user-defined spectrum region of\ninterest (SROI). A mental arithmetic experiment on a hybrid EEG-fNIRS was\nconducted to assess the efficiency of the method. Our proposed technique, STFT,\nand WT were applied on both biomedical signals to discover potential\noscillators that improve the discrimination between the task condition and its\nbaseline. The results show that DASAR provided the highest spectrum\ndifferentiation and it was the only method that could identify Mayer waves as\nnarrow-band artifacts at 97.4-97.5 mHz.\n