Adaptive signal processing for the real-time beat-by-beat detection of microvolt cardiac potentials.
Cardiovascular mortality continues to be the leading cause of death in the United
Kingdom, China and the United States. Many of these deaths occur suddenly, called sudden
cardiac death (SCD), with the number of these events estimated from these three Countries to be
over 1,500,000 annually.
In the vast majority of documented cases, the SCD is directly caused by ventricular
tachycardia (V1). Prediction of the presence of the VT is of great importance. It has been found
by using signal averaging (SA) techniques that the appearance of micro signals, called ventricular
late potentials (VLPs), is highly correlated with the appearance of VT or SCD. The VLPs are
about 0.1% - 1% of the size of the normal ECG in most patients and are masked by various noise
sources, so that they can not be seen from the standard electrocardiogram (ECG).
The SA techniques, depending largly on averaging many beats, can only detect the
microvolt signals that are strictly constant in duration, morphology and timing relative to the
QRS complex amongst the considerable amounts of noise which are present The main
disadvantage of the technique is that it cannot offer information from an individual beat, i.e.
variations among the beats and individual beat information are lost when averaging. This
information can be very important in the diagnosis of the development of many heart
abnormalities, particularly arrhythmias.
This thesis describes various techniques that have been developed for a real-time
processing system, in which the system can detect VLPs at the body surface with beat-to-beat
variations. One of the most important techniques is the use of adaptive filters to reduce the most
disruptive noise -random noise.
Clinical investigations have been carried out based on 14 normal and 20 abnormal
pathological subjects to produce reproducible results on the developed system. The results show
that the system can produce much more information than SA techniques for the prediction of VT.