A wavelet-based matlab suite for the analysis of EEG recordings

Introduction: Analyzing the electrical signals from the brain, electroencephalography (EEG), to examine the physiological and psychological well-being of an individual is common practice in psychophysiology. Often, this is accomplished by creating event-related potentials (ERPs) from multiple EEG recordings or by transforming the EEG recordings from the time domain to the frequency domain via Fourier transforms, typically using software for either method. Both methods reveal useful information that is typically inaccessible in the original recording. However, inherent deficiencies with both methods and the software used to execute them do exist. Many software programs obscure the routines by which the ERPs and Fourier transforms are generated. This lack of transparency complicates troubleshooting and innovation. In addition, exotic or novel methods of analyzing EEGs, such as the wavelet transforms, are not included in most software. In terms of the methods themselves, ERPs are by necessity a construct of many EEG recordings. In the instances in which few EEG recordings exist, an ERP is less effective as an analysis tool. The Fourier transform destroys potentially useful time information present in the original EEG recording. To rectify these shortcomings, an open Matlab program suite was devised using wavelet transform methods. Three sets of data were analyzed using the suite in order to determine usefulness and efficacy.;Methods: Three separate sets of data were obtained. Each set of data contained multiple groups exhibiting various psychological or physiological disorders (PTSD, Parkinson's disease, ADHD, etc.). The members of the groups participated in tasks designed to evoke a particular type of psychophysiological response. All tasks were carried out while the participants' EEGs were recorded using a 128 channel EGI system. The resulting recordings were exported, and a custom Matlab program suite based on wavelet methods was used to analyze the recordings. Implementation of the code for the test cases was the primary focus, but the results were also interpreted using statistical methods and compared to current literature to attempt to place them within a known framework.;Results: The Matlab program suite was successfully implemented for all test cases. Each data set was successfully converted to a format compatible with the program suite, and the wavelet methods successfully generated data for all sets. In addition, statistical analysis revealed many interesting significant differences between groups within each data set. Nearly all results could be explained using current publications and theories in a meaningful way.;Conclusions: A program suite based on Matlab and using wavelet-based methods was successfully created and implemented in multiple data sets of psychophysiological data consisting of EEG recordings. The program suite generated data in an alternative approach to that of traditional analysis techniques. These results were also useful in determining differences in brain function between various psychological and physiological disorders in certain psychophysiological tasks. This suggests that such a solution to the intrinsic problems with current EEG analysis techniques may be viable.