Blind source separation for selective tissue motion measurement in ultrasonic imaging

Multi-dimensional tissue motion measurement is of significant clinical relevance, yet conventional Doppler approaches to tissue velocity estimation in ultrasonic imaging are prone to error at large Doppler angles. Despite various correction methods and alternative approaches, no particular method has emerged as the premier technique for accurate and efficient multi-dimensional tissue motion measurement in clinical application. In addition to error inherent to velocity estimation schemes themselves, further velocity error is introduced by incomplete separation of signals corresponding to moving tissue structures of interest. Thorough signal separation is not necessarily possible via frequency-domain operations, as the signal of interest may contain frequency bands common to undesirable signal components. Rather, regression filters can be used to separate signals in the time-domain. Ideally, regression filters are implemented adaptively to automatically filter each data ensemble uniquely.; Blind Source Separation (BSS) is a method for enhanced adaptive signal segregation and multi-dimensional velocity estimation in ultrasonic imaging. BSS decomposes data ensembles into basis functions spanning orthogonal or independent signal components. In application to wall filtering for blood velocity estimation, BSS-derived basis functions effectively segregate vessel wall, blood, and noise signal components for subsequent blood velocity measurement without corruption from vessel wall and noise signals. Similar BSS methods are employed for adaptive filtering in Acoustic Radiation Force Impulse (ARFI) vascular imaging. Following BSS clutter filtering, small axial blood flow components at large Doppler angles are extracted from noisy axial velocity profiles with adaptive BSS filtering. BSS is further employed to reduce noise caused by jitter and physiological motion in measured ARFI-induced tissue displacement profiles. Finally, BSS is employed for complex phase clutter rejection to visualize ARFI-induced streaming in fluid filled cysts. In addition to adaptive filtering, BSS is directly capable of multi-dimensional tissue motion measurement via the novel technique, Blind Source Separation-Based Velocity Estimation (BSSVE). The method is demonstrated using simulated and clinical carotid artery data gathered from healthy, adult, male and female volunteers.; This dissertation supports the hypothesis, BSS is uniquely effective for adaptive filtering and multi-dimensional tissue velocity measurement, offering a novel approach to enhanced selective tissue motion measurement in ultrasonic imaging.