Hilbert-Huang Transform and Welch's Method for Motor imagery based Brain Computer Interface

Hilbert-Huang Transform and Welch's Method for Motor imagery based Brain Computer Interface

Omar Trigui (Advanced Technologies for Medicine and Signals ‘ATMS', ENIS, Sfax University, Sfax, Tunisia), Wassim Zouch (King Abdulaziz University (KAU) Jeddah, Saudi Arabia & Advanced Technologies for Medicine and Signals ‘ATMS', ENIS, Sfax University, Sfax, Tunisia) and Mohamed Ben Messaoud (Advanced Technologies for Medicine and Signals ‘ATMS', ENIS, Sfax University, Sfax, Tunisia)
DOI: 10.4018/IJCINI.2017070104
OnDemand PDF Download:
No Current Special Offers


The features extraction is the main step in a Brain-Computer Interface (BCI) design. Its goal is to create features easy to be interpreted in order to produce the most accurate control commands. For this end, these features must include all the original signal characteristics. The generated brain's signals' non-stationary and nonlinearity constitute a limitation to the improvement of the performances of systems based on traditional signal processing such as Fourier Transform. This work deals with the comparison of features extraction between Hilbert-Huang Transform (HHT) and Welch's method for Power Spectral Density estimation (PSD) then on the creation of an adaptive method combining the two. The parameters optimization of each method is firstly performed to reach the best classification accuracy rate. The study shows that the PSD estimation is sensitive to the parametric variation whereas the HHT method is mainly robust. The classification results show that an adaptive joint method can reach 90% of accuracy rate for a mental activity period of 1s.
Article Preview

1. Introduction

The BCI objective is to create a new communication or control channel between people and machines using only the brain activity. To achieve this target, the system must be able to discriminate among a set of different mental states. This technology is mainly promising for people suffering from severe neuromuscular disorders such as spinal cord injury, amyotrophic lateral sclerosis and brain stream stroke. Moreover, BCI can offer new interface device solutions for the control of video games. Users will be able to control objects or characters in the game through their mental activities and now, with the development of the new connected car services (Joo et al., 2015), BCIs can offer a promising solution to improve the navigation security.

Thanks to its low cost, non-invasiveness, portability, ease to use and high temporal resolution the EEG technology have been widely used in the field of BCIs (Saeid and Chambers, 2007). Actually, to express his intents, the subject must produce a specific brain electrical activity. Recently, many researchers have turned to the motor imagery based BCI. The motor imagery is the mental repetition of motor actions without apparent movements. It can stimulate a large number of cortical areas in the same way that the preparation or the execution of the real movement does (Qiu et al., 2015). This leads to the creation of local event-related changes in EEG spectra. The Event-Related Desynchronization (ERD) is a decrease in the amplitude of the signal power over the premotor and primary sensorimotor areas (Pfurtscheller, and Aranibar, 1977) (Chatrian et al., 1959). Likewise, the Event-Related Synchronization (ERS) is an increase in the amplitude of the signal power over the ipsilateral side due to the end of the motor tasks (Pfurtscheller et al., 1996) (Pfurtscheller et al., 1998) (Stancák and Pfurtscheller, 1996). These neurophysiological phenomena are basically useful to make difference between varieties of mental tasks. Many studies (Pfurtscheller, and Lopez da Silva, 1999) (Bai et al., 2005) (Huang et al., 2011) confirmed that motor imagery of left and right hand movements can be detected with the EEG signal from the channels C3 and C4 according to the 10/20 international system.

The signal processing entity of a BCI is usually composed of three successive steps. First, a pretreatment step to improve the signal quality. Then, a feature extraction one which converts the EEG signal to a more simple and useful one and finally a classification step to estimate subject’s mental state and convert signal into device commands according to the application.

The feature extraction step is the most interesting since it has the greatest effect on the classification accuracy rate which represents the ability of the system to precisely identify the brain states (Avilés-Cruz et al., 2016). In fact, although the classifier has a high ability to discriminate between the different classes, an imprecise characterisation can weaken its performances.

The features extraction technique based on the PSD estimation has received a great deal of attention in many EEG researches especially in motor imagery based BCI (Bashashati et al., 2007). For instance, some works focus on the comparison of this method with other ones such as the work of (Park et al., 2013) which studies the robustness of this technique facing small changes in electrode locations. (Diez et al., 2008) examine some alternatives of this method by studying the performance of the system using parametric methods for PSD estimation like Burg’s method and non-parametric method like Welch’s method. Furthermore, Welch’s method has been used for features extraction in a combination of a BCI with the intelligence of robots (Galán et al., 2008). Also, it was used in the design of a hybrid BCI with the parallel usage of the EEG and the electromyographic (EMG) activity (Leeb et al., 2011). In addition, this method was used in several researches which have emphasized on the comparison between the performances of different classifiers (Cincotti et al., 2003). (Djemili et al., 2014) have compared three spectral estimation methods used for the design of BCI systems. The standard periodogram, the Welch periodogram, and the logarithm of band powers have been used to characterize the right and the left-hand movements. Results show that Welch method gives the best precision levels.

Complete Article List

Search this Journal:
Volume 16: 1 Issue (2022)
Volume 15: 4 Issues (2021)
Volume 14: 4 Issues (2020)
Volume 13: 4 Issues (2019)
Volume 12: 4 Issues (2018)
Volume 11: 4 Issues (2017)
Volume 10: 4 Issues (2016)
Volume 9: 4 Issues (2015)
Volume 8: 4 Issues (2014)
Volume 7: 4 Issues (2013)
Volume 6: 4 Issues (2012)
Volume 5: 4 Issues (2011)
Volume 4: 4 Issues (2010)
Volume 3: 4 Issues (2009)
Volume 2: 4 Issues (2008)
Volume 1: 4 Issues (2007)
View Complete Journal Contents Listing