Extracting Information from Sonoelastograhic Images

Benjamin Castaneda Aphan

Professor Kevin Parker

CSB 426

Abstract

This thesis focuses on the implementation of image processing tools to extract information from images acquired with sonoelastography and crawling wave (CrW) sonoelastography. These algorithms enhance the quality of the images; extract location and size information of discrete lesions; and provide viscoelastic properties of the imaged tissue. Among the implemented tools, a semi-automated segmentation algorithm to measure discrete lesions in sonoelastographic images is proposed and evaluated through simulations, and experiments ex vivo and in vivo. The algorithm reduces variability and processing time in the measurements while keeping results comparable to manual segmentation. A second algorithm to process CrW images is implemented to extract shear velocity information from homogenous tissues. This correlation-based algorithm is successfully applied to the measurement of viscoelastic properties of human prostate ex vivo.  Finally, motion filtering and slow time processing are introduced to enhance the quality of the CrW images by exploiting their temporal and spatial harmonic properties. The proposed tools are applied to two important clinical applications: Prostate cancer detection and measurement of thermally ablated lesions in liver. In the former application, sonoelastography has an accuracy of over 80% for finding tumors larger than 4 mm in diameter, both in vivo and ex vivo, and slightly underestimates their volumes. CrW sonoelastography estimates the shear velocities of cancerous and normal prostate tissue as 4.75±0.97 m/s and 3.26±0.87 m/s, respectively. In the latter clinical application, results suggest that sonoelastography has the potential to be used as a complementary technique to conventional ultrasound for monitoring thermal ablation and follow-up imaging.