Date of Award
Fall 1980
Document Type
Dissertation - Restricted
Degree Name
Doctor of Philosophy (PhD)
Department
Biomedical Engineering
First Advisor
Jodat, Ronald W.
Second Advisor
Spurr, Gerald B.
Third Advisor
Milbrath, John
Abstract
The research described here relates to the formulation of new techniques of pattern recognition for application to medical thermography as applied to the detection of breast cancer in women. Breast cancer is the leading cause of cancer death in women with over 35,000 deaths annually in the U.S. alone. Clinical experience has demonstrated the efficacy of early detection in breast cancer with cure rates exceeding 90% in the early stages but deteriorating to less than 5O% in the more advanced stages. Medical thermography, which measures and interprets skin thermal patterns, is potentially an ideal detection technique for breast cancer. Unlike X-Ray techniques, it is completely passive measuring only the normal infrared radiation emanating from the body at room temperatures. It is also simple to execute with none of the operator skill requirements of X-Ray or ultrasonics. This research project has also demonstrated that thermographic instrumentation may be very cost-effective and therefore applicable to mass screening programs. With all of these real and potential advantages, however, the clinical results obtained with medical thermography have been disappointing. Most thermographic diagnosis has been based on manual interpretation of infrared photographs. Such interpretation has been severely handicapped by the lack of a standardized set of interpretation criteria. More recently, the advent of "computerized" thermography has introduced consistent interpretation criteria using statistical pattern recognition analysis. Even here, however, the improvement in diagnostic accuracy fell short of expectations. The end result is that medical thermography in its primary medical application of breast cancer detection has still not proven clinically useful...