Blind Source Separation Techniques Based Eye Blinks Rejection in EEG Signals

Automatic Extraction System for Common Artifacts in EEG Signals Based on Evolutionary Stone’s BSS Algorithm

“…ICA algorithms have an inherent disadvantage such as (i) source ambiguity, (ii) undetermined variances of the components, and (iii) the performance of ICA being decreased when the dataset is small and with large dataset the redundancy case is not sufficient to recover the independent components [2]. Stone's BSS is used instead of ICA due to these limitations [17][18][19]; it was first motivated by Stone [20]. Many researchers try to discuss and modify it to enhance the separation process [17,[21][22][23]].…”

Enhancement of Source Separation Based on Efficient Stone's BSS Algorithm

“…PSO and GA with a new fusion of fitness function are proposed in [10] to solve BSS. GA is used to tune the half-life parameters of Stone's BSS algorithm to enhance the separation process as explained in [3] and its used successfully to recover the original sources also Stone's BSS algorithm is proved to be the best in EEG signal analysis and separation [4]. Electrooculogram (EOG) artifact and power line noise interface are extracted successfully by Stone's BSS algorithm as explained in [5].…”

“…without or with very little knowledge of the original sources or the mixing process). Stone's BSS is a second-order statistic algorithm was proposed by Stone [1,2], it is a batch algorithm with low complexity and it is better than independent component analysis algorithms (ICA) [3][4][5][6]. Many researchers attempted to develop and discuss Stone's BSS [2,[7][8][9]], Stone's BSS is used for linear mixture, convolutive mixture [2] and for nonlinear mixtures [7].…”

“…Due to these problems, the original Stone's BSS is used to generate the initial population to overcome the 1st problem; the 2nd problem is solved by decreasing the number of population size. Many researchers have been used BSS techniques to separate both super-Gaussian and sub-Gaussian signal [3][4][5][6]11]. In EEG signal analysis eye blinking artifact is a super-Gaussian artifact and the power line noise is a sub-Gaussian interface [12].…”

“…(3) The performance of ICA is decreased when the dataset is small and with large dataset the redundancy case is not sufficient to recover the independent components [15]. Stone's BSS is used instead of ICA due to these limitations [3,4,16]. In this paper, an efficient Stone's BSS algorithm is proposed to separate both super-Gaussian and sub-Gaussian signals from their mixtures based on the joint between BSS algorithms and genetic algorithm.…”