Detection of electrocardiographic changes in partial epileptic patients using local binary pattern based composite...

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Clinical Neurology

Background:

• Epilepsy diagnosis often relies on clinical observation and electroencephalography (EEG).
• Electrocardiogram (ECG) signals contain physiological information that may correlate with neurological conditions like epilepsy.
• Existing methods for ECG-based epilepsy detection have limitations in accuracy and feature representation.

Purpose of the Study:

• To propose a novel composite feature set for enhanced detection of partial epileptic seizures from ECG signals.
• To evaluate the effectiveness of the proposed feature set using machine learning classifiers.
• To compare the performance of the new method against existing techniques for ECG-based epilepsy classification.

Main Methods:

• A composite feature set was developed by combining Local Binary Pattern (LBP) statistical features with statistical features from the original ECG signal.
• Two classifiers, Support Vector Machine (SVM) and a bagged ensemble of decision trees, were employed to assess the discriminating power of the features.
• Experiments were conducted on the publicly available MIT-BIH ECG dataset.

Main Results:

• The proposed composite feature set demonstrated superior performance compared to conventional histogram-based LBP features.
• The method achieved higher classification accuracy in distinguishing between normal and partial epileptic ECG beats.
• The combination of LBP and original signal features enhanced the discriminative ability for epilepsy detection.

Conclusions:

• The novel composite feature set offers a promising approach for improving the accuracy of ECG-based partial epilepsy detection.
• The proposed method outperforms existing literature approaches for classifying epileptic beats in ECG.
• This technique has the potential to aid in the non-invasive diagnosis and monitoring of epilepsy.