Abstract
With the rapid innovation in the field of healthcare, various biomedical signals, namely, electrocardiogram (ECG), electroencephalogram (EEG), electromyogram (EMG), play a crucial role for accurate measurement of various diseases such as cardiovascular diseases, brain disorders, etc. In the present work, an efficient method based on empirical mode decomposition (EMD) has been proposed to detect the epileptic activity. The present study is composed of three parts. In the first part, EMD is used to decompose the EEG signal into a set of amplitude modulated and frequency modulated components, referred to as intrinsic mode functions (IMFs). In the second part, features such as standard deviation, kurtosis, and Hjorth parameters have been extracted from various IMFs. In the last stage, the features are employed as inputs to support vector machine classifier for classification between non-seizure and seizure EEG signals. The simulation results show that the proposed scheme has attained better classification accuracy when compared to existing state-of-the-art methods.
TopIntroduction
In today’s world, more and more people want to live a physically, emotionally and mentally prosperous life and are thus, concerned about their own health and well-being in all aspects of their body and life style. Technology plays an important role in the diagnosis, monitoring, interpretation and characterization of an individual’s health and well-being. Not only does it help the people to analyse the simple status of their weight or blood pressure but also, with more and more advancements, it helps understand the complex activities of heart, brain etc. Thus, an individual’s health can be addressed as a whole, which includes the simplest to the most complex functioning of various organs(Turnip et al., 2018, Singhal at al., 2020).
Brain is one such organ with complex functioning. It has been studied to generate electrical signals or an impulse because of the movement of charged molecules called ions through voltage gated channels in the plasma membrane of neurons. When neurons are at rest, a charge difference is maintained (inside: negative, outside: positive) across the membrane of the cell (resting potential), majorly because of the transport by Na+- K+ pumps. In presence of stimuli, the sodium channels open, which lead to the depolarization of the membrane and change in the polarity across it (action potential). In order to repolarize the membrane, the potassium channels open, thus restoring the resting potential. Thus, an impulse travels down the nerve cell and gets converted into a chemical signal in the end. These chemicals are called neurotransmitters which pass the messages from one neuron to the other.
With such complex functioning of the brain, there exist some disorders that affect the brain and its functioning. These include simple problems like headaches, dizziness, sleeping problem, change in behavior etc. and also major problems like brain activity during injury, heart or liver transplant, brain tumor, seizures, epilepsy, brain dead, etc. Various studies to understand these activities of the brain and the development of related technology were conducted between 1875-1900. Brain waves or neural oscillations produced due to synchronized electric impulses of millions of neurons were also observed. Many years later, it was understood that five different waveforms are generated in the brain, each producing a different signal. These include alpha, beta, theta, delta, and gamma waves. Their frequencies and role in differentiating between the normal activities of the brain from the abnormal one are listed below.Finally, in 1924, after sufficient studies about brain functioning, Hans Berger, recorded the first human EEG. The EEG signalmeasures the voltage fluctuations occurring within the neuron of the brain. It is used to monitor the abnormal electrical activity in the brain by placing multiple electrodes across the scalp.
Table 1.
Comparison of different waveforms present in the brain
| Band | Frequency(Hz) | Location | Normally |
| Delta | <4 | Frontally in adults, posteriorly in children, high amplitude waves | Adult slow-wave sleep, it has been found during some continuous-attention tasks |
| Theta | 4-7 | Found in locations not related to task at hand | Higher in young children, drowsiness in adults and teens |
| Alpha | 8-15 | Posterior regions of head, both sides higher in amplitude on dominant mode | Relaxed/reflecting, closing the eyes, associated with inhibition control, inhibitory activity in different locations across the brain |
| Beta | 16-31 | Both sides, symmetrical distribution, most evident frontally, low amplitude waves | Active thinking, focus, high alert, anxious |
| Gamma | >32 | Somatosensory cortex | Display during cross-modal sensory processing, shown during short-term memory matching of recognized objects, sounds or tactile sensation |
Key Terms in this Chapter
Empirical Mode Decomposition (EMD): It is an adaptive method used to decompose non-linear and non-stationary time-series signal into a set of mono-component signals known as intrinsic mode functions (IMFs).
Hjorth Parameters (HP): time-domain parameters commonly used in the analysis of biomedical signals for feature extraction.
Quasi-Stationarity Signal (QSS): A type of non-stationary signal which can be represented as stationary in the local time frames.
Action Potential (AP): A mechanism through which nerve cells communicate and conduct information.
Electroencephalogram (EEG): It is non-invasive procedure which is used to record the electrical activity of the human brain.
Support Vector Machine (SVM): It is a supervised machine learning method used for classification and regression analysis.
Cubic Spline Interpolation (CSI): A type of interpolation technique used to reduce the effect of over fitting.
Intrinsic Mode Functions (IMFs): It is a collection of basic modes obtained from the application of EMD method.
Hilbert Transform (HT): It derives the analytic representation of the real-valued signal.