Electrical and Systems Engineering
Date of Award
Doctor of Philosophy (PhD)
Chair and Committee
Signals associated with biological activity in the human body can be of great value in clinical and security applications. Since direct measurements of critical biological activity are often difficult to acquire noninvasively, many biological signals are measured from the surface of the skin. This simplifies the signal acquisition, but complicates post processing tasks. Modeling these signals using the underlying physics may not be accurate due to the inherent complexities of the human body. The appropriate use of such models depends on the application of interest. Models developed in this dissertation are motivated by underlying physiology and physics, and are capable of expressing a wide range of signal variability without explicitly invoking physical quantities. An approach for the processing of biological signals is developed using graphical models. Graphical models describe conditional dependence between random variables on a graph. When the graph is a tree, efficient algorithms exist to compute sum-marginals or max-marginals of the joint distribution. Some of the variables correspond to the measured signal, while others may represent the hidden internal dynamics that generate the observed data. Three levels of hidden dynamics are outlined, which enable models to be constructed that track internal dynamics on differing time scales. Expectation maximization algorithms are used to compute parameter estimates. Experimental results of this approach are presented for a novel method of recording bio-mechanical activity using a Laser Doppler Vibrometer. The LDV measures surface velocity on the basis of the Doppler shift. This device is targeted on the neck overlying the carotid artery, and the proximity of the carotid to the skin results in a strong signal. Vibrations and movements from within the carotid are transmitted to the surface of the skin, where they are sensed by the LDV. Changes in the size of the carotid due to variations in blood pressure are sensed at the skin surface. In addition, breathing activity may be inferred from the LDV signal. Individualized models are evaluated systematically on LDV data sets that were acquired under resting conditions on multiple occasions. Model fit is evaluated both within and across recording sessions. Model parameters are interpreted in terms of the underlying physiology. Pressure wave physics in a series of elastic tubes is presented to explore the underlying physics of blood flow in the carotid. Mechanical movements of the carotid walls are related to the underlying pressure, and therefore the cardiovascular activity of the heart and vasculature. This analysis motivates a model that can be estimated from experimental data. Resulting models are interpreted for the LDV signal. The graphical models are applied to the problem of identity verification using the LDV signal. Identity verification is an important problem in which the claimed identity is either accepted or rejected by an automated system. The system design that is used is based on a loglikelihood ratio test using models that are trained during an enrollment phase. A score is computed and compared to a threshold. Performance is given in the form of False Nonmatch and False Match empirical error rates as a function of the threshold. Confidence intervals are computed that take into account correlations between the system decisions.
Kaplan, Alan, "Information Processing for Biological Signals: Application to Laser Doppler Vibrometry" (2011). All Theses and Dissertations (ETDs). 890.
Permanent URL: http://dx.doi.org/10.7936/K7Q81B4T