Date of Award

Spring 1970

Document Type

Dissertation

Degree Name

Doctor of Engineering Science in Mechanical Engineering

Department

Mechanical Engineering

First Advisor

Martin J. Levy

Second Advisor

Herbert Barkan

Third Advisor

Russell W. Brancato

Fourth Advisor

Eugene H. Smithberg

Fifth Advisor

Rong-Yaw Chen

Abstract

The general problem of identifying significant characteristics of a system by analyzing the properties of a signal emitted by that system is common to many disciplines, biological as well as physical. In general, achievement of a satisfactory solution depends on the capability of assigning membership of candidate signals to one of a number of mutually exclusive categories:

In this study, a new technique of processing the standard electrocardiograms from human subjects has been developed. This technique employs a set of non-linear transformations which enable the assignment of electrocardiograms into one of two Mutually exclusive categories. The first category is that of patients with occult coronary artery disease; the second category is that of subjects free from coronary artery disease.

An introduction to the conventional means for assessing the standard electrocardiogram is presented, and the limitations noted. The new method is then presented which consists in large part of combining non-linear signal processing with a topological re-orientation of conventional cartesian coordinates to yield a multi-vector space which offers an optimum degree of visual perceptibility. Moreover, in this new domain, the transformed EKG tracings taken from subjects free from coronary artery disease inhabit a closed area, represented by the annular space bounded by two tangent circles. It is then shown that EKG tracings taken from subjects with occult coronary artery disease exhibit patterns which protrude beyond the circular boundaries.

The mathematical transformations are shown to be non-linear and non-analytic in terms of satisfying the Cauchy-Riemann conditions. The assumptions and constraints of the transformations are explored and practical considerations are examined.

Preliminary results obtained to date from over 100 medical cases are presented and analyzed. The tentative findings reveal a high degree of promise for the detection of asymptomatic coronary candidates well in advance of myocardial infarction. Finally, further applications to other medical areas are explored including a discussion of potential applicability to the fields of pharmacology and neurology.

Share

COinS