Signal processing to enhance ultrasonic tissue characterization in diagnostic medical imaging

Abstract

The goal of this presentation is to lay the background for identifying potential new contributions to medical diagnosis with ultrasonic imaging that might be provided by advanced signal-processing techniques. The focus of this talk is on ultrasonic tissue characterization, an approach that complements and extends presently available high-resolution imaging. Regional estimates of intrinsic material properties of normal and pathologically altered tissue are obtained from local measurements of attenuation and backscatter derived from the same signals that are used to form the ultrasonic images. Examples are drawn primarily from echocardiography, in part because imaging and characterizing the beating heart present significant challenges. Current echocardiographic images, frequently acquired with harmonic mode (nonlinear) imaging, exhibit substantial resolution of anatomical detail. Nevertheless, the estimates provided by tissue characterization of alterations in tissue properties arising from specific pathologies offer the promise of elevating echocardiographic imaging to a new plane. Among the challenges to be met in order to accomplish this goal are properly accounting for the intrinsic anisotropy of the heart and overcoming limitations that arise as a result of phase and amplitude aberrations associated with propagation through an intrinsically inhomogeneous medium. [Work supported by NIH R37HL40302 and R01 HL72761.]