Controlling Prosthetic Limb Movements Using EEG Signals

Controlling Prosthetic Limb Movements Using EEG Signals

V. V. Ramalingam (SRM University, India), Mohan S. (Al Yamamah University, Saudi Arabia), V. Sugumaran (VIT University, India), Vani V. (Al Yamamah University, Saudi Arabia) and B. Rebecca Jeya Vadhanam (SRM University, India)
DOI: 10.4018/978-1-5225-0889-2.ch008
OnDemand PDF Download:
No Current Special Offers


This chapter focuses on replacing natural arms with artificial arms with movement controlled by EEG signals. The selected features were classified using C4.5 decision tree algorithm, best first decision tree algorithm, Naïve Bayes algorithm, Bayes net algorithm, K star algorithm and ripple down rule learner algorithm. The results of statistical and histogram features are discussed and conclusions of the study are presented.
Chapter Preview

1. Introduction

The loss of human limb is a major issue that intensely limits the everyday capabilities and interaction of the persons. There can be two types of signals that are of direct use for the above purpose: EMG and EEG. EMG signals are available in the muscles and they contain a large amount of information for the purpose of limb movements. However, there are many instances where the subject loses most part of the limb. In such cases, the EMG signals that are available near the affected area (shoulder, upper arm) may not be of great use. Moreover, EMG signals are secondary signals, whereas EEG signals are primary signals. Since the EEG signals originate from brain activities, the characteristics remain almost same irrespective of the extent of amputation. This gives a feeling that EEG signal is better candidate for controlling movements of artificial limbs. It is not fully true because of the fact that the EEG signals are a product of some thought process. This complicates the decoding process of EEG signals. Now, the challenge is to decode effectively the information buried inside the EEG signals. A review of the techniques brings out merits and demerits. The need and scope of the present work also evolve from this. The main objective of the study is used to replace the natural arms with an artificial arm with movements of different right hand limb movements like finger open (fopen), finger close (fclose), wrist clockwise (wcw) and wrist counterclockwise (wccw) from EEG signals. These EEG signals can be used to build a model to control the prosthetic limb movements. Cluster analysis is a multivariate statistical analysis method; it is a statistical classification approach that groups signals into different fault categories on the basis of the similarity of the characteristics or features they possess. It seeks to minimize within-group variance on the one hand and maximize between-group variance on the other. The result of cluster analysis is division into a number of groups with homogeneous contents. Multilayer Perceptron Neural Network (MLPNN) architecture was employed to classify the EEG signals. Three sets set-A, set-D and set-E of EEG signals were used for classification. Lyapunov exponents were extracted from the EEG signals given as the inputs to the MLPNNs. Finally, the features were trained with Levenberg-Marquadt algorithm and achieved good classification accuracy (Elif Derya Ubeyli, 2009). Che Wan Fadzal et al. (2012) reported that the Power Spectral Density (PSD) was the well suited method to distinguish right and left hand writing movements using EEG signals. The controlling process of prosthetic limb movements based on surface EMG signals extracted from remnant muscles are the promising ones in the analysis of EMG signals. There were three feature extraction techniques, namely autoregressive coefficients, mean frequency and EMG histogram used in the study. The combined features of mean frequency and EMG histogram were given as the input to neural networks classifier. Hence, it is noticed from this study that the EMG histogram feature vector performed well for the classification of prosthetic limb movements (Aishwarya et al., 2013). Adeli et al. (2003) reported that the wavelet was an effective time–frequency analysis tool for analyzing EEG signals and most capable technique to extract features from EEG signals. The main idea of time series modeling is to fit the waveform data to a parametric time series model and extract features based on this parametric model. There are two popular mathematical models, namely, Auto-Regressive (AR) model and the Auto-Regressive Moving Average (ARMA) model. AR model is established by the time difference and EEG amplitude. In practice, however, application of the AR model or ARMA model is difficult due to the complexity in modeling, especially the need to determine the order of the model (Andrew, 2006). Least Square Support Vector Machine (LS-SVMs) was proposed by (Elif Derya Ubeyli, 2010). To classify normal and epilepsy patients during epileptic seizures, for feature extraction, spectral analysis of the EEG signal was carried out with three model-based methods namely, Burg autoregressive-AR, moving average – MA, Yule- walker autoregressive moving average-ARMA methods. The author has proved that the Burg AR coefficients were the best features to represent the characteristics of EEG signals. Shiliang et al. (2007) systematically evaluated the performance of the three ensemble methods for EEG signal classification of mental imaginary tasks. K-nearest-neighbor, decision tree and support vector machine were used as the classifiers and the experiments are carried out upon real EEG recordings from the mental imaginary task. Artificial Intelligence (AI) techniques have been increasingly applied to control prosthetic limb movements and have shown improved performance over conventional approaches. Numerous attempts have been made to improve the accuracy and efficiency of prosthetic limb movements by employing AI techniques. Various supervised learning techniques have been applied for controlling prosthetic limb movements. Ubeyli et al. (2007) used a diverse and composite input features of EEG signals obtained by the eigenvectors. The classification was performed on diverse features (modified mixture of expert-MME) and composite feature (mixture of experts-ME) with five data set (set-A, set-B, set-C, set-D, and set-E). The result demonstrated that the MME trained on diverse features have achieved a higher level of classification accuracy than ME. Ocak (2008) new scheme was presented to classify the EEG signals. Features were extracted using fourth-level Wavelet Packet Decomposition (WPD). Genetic Algorithm (GA) was used for feature selection and identified the best performing features to form the optimal feature subset. The approximate entropy values were derived as the feature vector and a Learning Vector Quantization (LVQ) was used as the classifier to attain the best classification accuracy for the normal and epileptic epoch. A review of recent developments in applications of ML techniques for controlling prosthetic limb movements was given by Bhattacharyya et al. (2015), Ericka et al. (2015), Monalisa et al. (2015), Enrique et al., (2015), and Siddique et al. (2003). To make the training and testing of the techniques (classifiers) more effective, feature selection was employed using Decision Tree (DT) (Suykens et al, 2003).

Complete Chapter List

Search this Book: