| Summary: | 博士 === 國立清華大學 === 生醫工程與環境科學系 === 101 === The primary goal of this work is to exploit the physical characteristics of feature type for better classifying breast tumors using ultrasound. The main topics can be divided to three certain parts. In chapter 2, ultrasound B-scan-based morphological and texture analysis and Nakagami parametric imaging were proposed to characterize breast tumors. These feature categories of ultrasound tissue characterization supplied information on different physical characteristics of breast tumors, by combining the above methods was expected to provide more clues for classifying breast tumors. The empirical results indicated that the combination of morphological-feature parameter (e.g., standard deviation of the shortest distance), texture feature (e.g., variance), and the Nakagami parameter resulted in the specificity and sensitivity both exceeded 88%, and the area under ROC curve of 0.95. In chapter 3, the feasibility of applying the elasticity imaging method to Nakagami imaging was investigated for visualizing the local redistributions of scatterers in a scattering medium with different stiffnesses. The preliminary results show the concept of the elasticity and scatterer characterizations being functionally complementary in classifying breast tumors. Consequently, the sequential Nakagami image frames obtained from different strain conditions would simply represent the relative tissue stiffness. In chapter 4, the use of a strain-compounding technique with Nakagami imaging was presented to identify breast lesions, which was regarded as strain-compounding in the Nakagami domain to provide a new parameter associated with the scatterers and stiffness of tissues. Combining information from multiple Nakagami images obtained under different strain conditions can be useful to improve the ability to interpret the characteristics of breast tumors. Potential applications of these proposed imaging techniques include extending to the three-dimensional ultrasound for classifying different stages and grades of breast tumors, and assisting abdominal or musculoskeletal ultrasound for detecting characteristic symptoms and abnormalities.
|
|---|
