Biometric Identification From ECG Signals Using Fourier Decomposition and Machine Learning

Abstract

This article investigates biometric identification systems based on electrocardiogram (ECG) signals and their intrasubject and intrasession validity. We develop an efficient algorithm using Fourier decomposition method (FDM) and phase transform (PT). First, the ECG signal is divided into frames consisting of one or more beats. These frames capture both interbeat and intrabeat variations. They are decomposed into a set of Fourier intrinsic band functions (FIBFs) using FDM and relevant features are extracted from them. In addition, PT has been used to highlight the intrinsic information hidden in the phase of ECG signals. The effects of variations in the size of the frame, the decomposition levels, and the number of sessions used for training and testing on the performance of the algorithm are analyzed. Random forest (RF), ensemble subspace discriminant (ESD), and support vector machine (SVM) are applied as classifiers to evaluate the performance on three datasets, MIT-BIH, ECG-ID and Check Your Biosignals Here Initiative (CYBHi), where MIT-BIH is acquired in an on-the-person setting and the other two are off-the-person datasets. The proposed method achieved identification accuracies of 91.07% for the CYBHi dataset, 97.92% for the MIT-BIH dataset, and 98.45% for the ECG-ID dataset, which are better than most of the existing state-of-the-art algorithms. © 1963-2012 IEEE.