System Optimization For Breast Sonography

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): This project aims to develop a first-principles approach to the design and evaluation of breast sonography specifically for cancer detection and malignant lesion discrimination. The approach is based on the Bayesian ideal observer model for breast sonography that reveals optimal strategies for image formation based on maximizing diagnostic information content for specific examinations. Our current approximation to the ideal observer strategy is to use coded-pulse excitation methods with a series of spatial Wiener filters applied to data from a fixed-focus, large-aperture array. Extensions proposed in this project include iterative Wiener filters to extend the method to highly nonstationary scattering media, and development of the ideal observer for pre-beamformed echo data that allows optimization of beamforming strategies for specific clinical tasks. Preliminary imaging results and studies involving psychophysical testing have shown that lesion visibility can be significantly improved over standard beamforming via dynamic receive focusing. We wish to develop this new imaging approach from the basic theory of image formation, through implementation of the most promising methods on a clinical instrument, and finally testing with tissue mimicking phantoms. These investigations may lead to new optimal techniques for breast imaging and a new perspective on how to design and evaluate all ultrasonic imaging techniques. We are developing a novel approach to ultrasonic breast cancer imaging that shows great promising for significant improvements in diagnosis. If successful, sonography will play a more prominent role in cancer screening using techniques that are highly effective, safe and low cost.