Date of Award
Spring 1995
Document Type
Thesis - Restricted
Degree Name
Master of Science (MS)
Department
Biomedical Engineering
Abstract
Ultrasound tissue characterization has been used to determine histopathologic information in organs such as the heart, eye and liver. Tissue characterization uses different signal processing algorithms to extract information from the ultrasound signals acquired from the tissue. At present, available screening modalities for prostate cancer are limited in their ability to detect and stage pathologic conditions. There is a need to improve the sensitivity and specificity of cancer screening techniques. In this particular study, a technique called cepstral analysis will be used to quantify prostate tissue properties. Scatterers in the tissue will cause reverberations in the ultrasound signals. By performing cepstral analysis, these reverberations will appear as peaks in the cepstrum at frequency inversely proportional to the distance between scatterers. It has been postulated that during pathologic conditions, scattering properties of a tissue may change, and these changes may produce a characteristic pattern or peak in the cepstrum. As a matter of fact, these changes can be used to distinguish the pathologic lesion, such as cancer from normal tissue. The cepstrum can be obtained by taking the Fourier Transform of the logarithm of the power spectral density of a signal. Autoregressive Moving Average modeling is used to find the power spectral density of the ultrasound signals, the reason is because it can produce a smooth spectral estimate with less fluctuation. Based on this study, cepstral analysis technique in the detection of cancers greater than 0.2cm3 in volume has a sensitivity of 66.58%, specificity of 92.84%, positive predictive value (PPV) of 58.94%, and negative predictive value (NPV) of 93.54%. The results of this study suggest that cepstral analysis of ultrasound tissue characterization of the prostate, is able to detect cancer foci confined to the prostate gland with a volume greater than 0.2cm3.
Recommended Citation
Hau, William Kongto, "Ultrasound Tissue Characterization of Prostate Cancer by Using Cepstral Analysis Technique" (1995). Master's Theses (1922-2009) Access restricted to Marquette Campus. 4781.
https://epublications.marquette.edu/theses/4781