Enhancing Biometric Identification Through EEG and ECG Fusion

Biosignals have gained popularity in biometric identification system studies due to their characteristics, which are difficult to duplicate due to liveliness property. Two biosignals commonly used in biometric identification systems today are electroencephalogram (EEG) and electrocardiogram (ECG) signals. This study aims to leverage the unique properties of EEG and ECG signals through their integration at the feature level using a stacking and convolution process within a convolutional neural network (CNN). The fusion of these two biosignals is expected to improve both the accuracy and security of biometric identification systems. The study employs the N&C-TEC dataset, consisting of EEG and ECG signals from 15 subjects, to evaluate the effectiveness of the proposed method. Through stratified cross-validation on training and testing datasets, the proposed method achieved a high accuracy rate of 99.46% for both sets, demonstrating the significant potential of multimodal biosignals in the biometric identification systems. The results highlight the advantages of combining EEG and ECG signals, addressing the limitations of unimodal biometric systems and providing a more robust, secure, and accurate identification method.