Date of Award

Summer 1999

Document Type

Thesis - Restricted

Degree Name

Master of Science (MS)


Biomedical Engineering

First Advisor

Ropella, Kristina M.

Second Advisor

Jeutter, Dean C.

Third Advisor

Prieto, Thomas E.


Electronic therapy has rapidly become the therapy of choice for many people who experience cardiac arrhythmias. Proper administration of therapy by implantable electronic anti-arrhythmic devices depends, in part, on the ability of the device to accurately differentiate various heart rhythms. The first generation of implantable cardioverters/defibrillators (ICD) relied on several heart rate criteria estimated from intracardiac electrograms for arrhythmia identification [1,2]. However, these rate-related algorithms misinterpret arrhythmias due to physiological overlap of heart rate between pathologic and non-pathologic rhythms [2,3]. In addition, these algorithms are highly sensitive to changing signal amplitude and morphology [4,5]. Newer generations of ICDs incorporate such criteria as RR interval stability [34], sustained rate duration criteria [34], and syntactic pattern recognition [34]. However, all of these arrhythmia detection schemes demonstrate a trade-off between sensitivity and specificity [34]. More recently proposed methods for arrhythmia interpretation include contextual analysis [1,2,4,6] and correlation techniques [7-9,35] that utilize time-domain characteristics of waveform amplitude and morphology. However, these methods require multiple lead systems and are sensitive to lead configuration and stability of lead placement [1,4,6]. As an alternative to time-domain analysis, frequency-domain analysis has been shown to be a reliable discriminator of pathological and non-pathological rhythms [10-12, 13,36]. Previously, Rope Ila et al. [11] have shown magnitude-squared coherence (MSC), a frequency-domain measure of the constancy of phase relationship (time delay) between two bipolar electrograms, completely differentiates fibrillatory from nonfibrillatory rhythms. In these previous studies, MSC estimation required two bipolar leads. However, newer generations of ICDs have a single transvenous lead with two sensing electrodes. Thus, the use of MSC in current devices would require implementation with a single lead rather than two leads. The main objectives of this thesis are to: evaluate whether or not MSC may be implemented using a single intracardiac lead with two electrode elements and whether such MSC measurements may differentiate fibrillatory from nonfibrillatory cardiac rhythms. We hypothesize that MSC between the bipolar signal (local activity) and one of the unipolar signals (global activity) that contributes to the bipolar signal would be high for nonfibrillatory ventricular rhythms (NVF) and relatively low for ventricular fibrillation (VF) [ 13]. • evaluate and compare a real-time MSC algorithm based on traditional FFT estimation with a real-time MSC algorithm based on a Moving OFT, and real-time Rate in its ability to differentiate fibrillatory from nonfibrillatory rhythms when using unipolar electrograms [8]. • investigate the feasibility of implementing a real-time MSC algorithm in an integrated microprocessor chip. This investigation addresses the concerns that the large amount of complex computation involved in frequency-domain analysis would limit its use in practical applications.



Restricted Access Item

Having trouble?