A robust parametric estimator for single-trial movement related brain potentials

Abstract

Current estimators for single-trial evoked potentials (EP's) require a signal-to-noise ratio (SNR) of 0 dB or better to obtain high quality estimations, yet many types of EP's suffer from substantially lower SNR's. This paper presents a robust-evoked-potential-estimator (REPE) facilitating high quality estimations of single movement related EP's with a relatively low SNR. The estimator is based on a standard ARX model, enhanced to support estimation under poor SNR conditions. The REPE was tested successfully on a computer simulated data set giving reliable single-trial estimations for the low SNR range of around -20 dB. THe REPE was also applied to experimental data, producing clear single-trial estimations of movement related brain signals recorded in a classic scenario of self-paced finger tapping experiment.

Optimization of wavelets for classification of movement-related cortical potentials generated by variation of force-related parameters

“…Many features have been proposed for the pattern recognition task in BCIs, such as parameters of autoregressive models (Cerutti et al, 1988;Lange and Inbar, 1996), Fourier transformation coefficients (Millan et al, 2002;Pineda et al, 2000), and wavelet transform coefficients (Demiralp et al, 1999;Thakor et al, 1993). The adopted classification method is based on a feature space previously proposed by Maitrot et al (2005), with the use of a more complex classification scheme and its optimization.…”

Spatial and Temporal Distribution Enhancement of Movement-Related Brain Macropotentials

“…On the other hand, the indubitable ability of the identification procedure to extract the EP or ER content from the single-sweep raw data has been widely demonstrated in the literature, also by the authors. In fact, it has been successfully applied to MRBM (10,11,12,25), evoked magnetoencephalogram (21), visual evoked potentials (VEP) (23), somatosensory evoked potentials (SEP) (24), and contemporary visual and somatosensory stimulation, together with VEP-SEP recordings (11,22). As already stated, the final goal of the present paper concerns the applicability of the ARX to SCD brain mapping and our results suggest which is the better order of application of the Laplacian and ARX procedures to the raw data in order to get a reliable and more easily readable single-sweep brain map.…”

Latency change estimation for evoked potentials: A comparison of algorithms

“…Other kinds of noise may also be present during EP signal acquisition. The overall signal to noise ratio (SNR) may be as low as 10dB (MATSON, 1988;VAZ and THAKOR, 1989) under some situations, and the SNR may be lower than ~0dB (LANGE and INBAR, 1996;LANGE et al, 1997). EP signals can also be contaminated by external interferences from sources such as power lines and disturbances due to movement of the recording electrode (PHIIAPS, 1996).…”