As a medical diagnostic technique, proposed by Willem Einthoven in the early 1900s, the electrocardiogram(ECG) has a relatively long and illustrious history. It has since been acknowledged as an indispensable tool in the detection and treatment of various cardiac disorders. More recently, the ECG has fulfilled a rather unlikely niche, as a purveyor of security and privacy in the form of a biometric modality. In this capacity, there are various implications and technical challenges. In order to address these obstacles, we propose novel signal processing techniques that seek to not only establish the status of the ECG as an indisputable fixture in biometric research, but also reinforce its versatile utility, such as in alleviating the resource consumption in certain communication networks.

Physiology of the Electrocardiogram

ECG signals reflect the variations in electrical potential of the heart over time. The change in voltage is due to the action potentials of cardiac cells. The electrical activity is initiated when the sinoatrial (SA) node, the pacemaker of the heart, depolarizes. This electrical signal then travels rhythmically until it reaches the atrioventricular (AV) node, which is responsible for delaying the conduction rate, to properly pump blood from the atria into the ventricles.

Fig. 1 shows the salient components of an ECG signal: the P wave, the QRS complex and the T wave, which together account for the sequential depolarization and repolarization of the heart.

Figure 1 - Main components of an ECG signal

 

The P wave describes the depolarization of the right and left atria. The amplitude of this wave is relatively small, because the atrial muscle mass is limited. The QRS complex corresponds to the largest wave, since it represents the depolarization of the right and left ventricles, being the heart chambers with substantial mass. Finally, the T wave depicts the ventricular repolarization. It has a smaller amplitude, compared to the QRS complex, and is usually observed 300 ms after this larger complex. However, its precise position depends on the heart rate, e.g., appearing closer to the QRS waves at rapid heart rates.

The heart rate of a normal sinus rhythm is 60-100 beats/min (bpm). However, this is highly dependent on emotional factors, such as stress, anxiety, and shock, as well as on cardiovascular activities, such as running and exercising.

ECG Signal Acquisition

One of the main problems in biometric signal processing is the high degree of noise and variations. In many cases, a reliable acquisition is only possible with sufficient knowledge of the spectral content, the dynamic range and other characteristics of not only the desired signal components, but also of the noise sources involved. This is so that the appropriate filters and quantizers can be accordingly constructed to extract the desired signals, and reject the noise sources.

Based on the salient characteristics of ECG signal components, the P wave is a lower-amplitude and lower-frequency signal, while the QRS complex exhibits larger-amplitude and higher-frequency variations. In addition, the following sources of noise and artifacts are relevant to ECG. The baseline wander, arguably one of most common artifacts, refers to a low-frequency interference in the ECG, which may be induced by cardiovascular activities. The amplitude change due to baseline wander can potentially exceed the QRS amplitude by several times, which can be highly problematic for accurate medical diagnoses based on the isoelectric line. While this distortion may exhibit higher frequencies, e.g. during strenuous exercise, its spectral content is typically limited to an interval below 1 Hz [2]. Thus, some type of low-pass filtering would be relevant to this scenario.

Another source of error is powerline interference, being 50 or 60 Hz depending on the geographical location, which occurs due to insufficient grounding or interferences from other equipments. Also present in practical ECG recordings are electrode motion artifacts, due to skin stretching which alters the impedance around the electrode. These artifacts are problematic since their spectral content, being 110 Hz, overlaps that of the desired signal components.

As well, there are inherent physiologically induced artifacts, viz., respiratory activity artifacts. The involved chest movements change the position of the heart and the lung conductivity, leading to not only variations in the heart rate, but also modifications of the beat morphology. Clearly, as in medical applications, an ECG-based biometric system needs to take into account all these various sources of error, using the appropriate preprocessing, e.g., filtering based on the specific spectral contents.