Current ultrasonic imaging technology results in images which, despite advances in transducer technology, still contain a high level of distortion. The ultrasonic image is obtained by measuring the reflections of an ultrasound wavelet transmitted from a transducer positioned on the skin. Depth variations in the acoustic impedance of subdermal tissues produce reflections which can be used to image tissue boundaries. Distortion in the lateral direction, perpendicular to the propagation path of the ultrasound, is due to diffraction spreading of the ultrasound beam and can be compensated by using phased arrays to focus and steer the main ultrasonic beam in the desired direction. Another significant source of distortion in ultrasonic images occurs on the axial direction, along the ultrasound propagation path, and is due to reverberation resulting from a convolution of the outgoing ultrasonic wavelet with the intervening tissue reflection profile. Most attempts at compensating for axial distortion have involved some form of deconvolution where the reflected signal is filtered by an inverse filter in an attempt to increase the axial resolution of the image. There are a number of problems inherent in the deconvolution approach which have been well-documented. These include unrealizable inverse filters due to non-minimum phase wavelets, a high sensitivity to noise, unrealistic assumptions about tissue and transducer characteristics, and a statistical stationarity requirement - implying that deconvolution does not readily lend itself to imaging time-varying tissue structures. The method proposed here addresses these problems by using adaptive system identification to measure ultrasonic images. The significance of this approach is that, unlike deconvolution methods, system identification makes no assumptions about the tissue or transducer characteristics, and is much less susceptible to noise. The system identification concept is also particularly well suited for time-varying tissue structures. A new system identification algorithm is described which is particularly well-suited for this task given its high insensitivity to measurement noise. Equipment for setting up an ultrasound imaging lab is sought. The equipment includes ultrasound transducers, a pulser/receiver for exciting the transducer and conditioning the received ultrasound, a water tank for holding the transducer and target, an assembly that allows very accurate control of transducer/target position, and a digital oscilloscope for doing high speed data acquisition and display. This research seeks to improve the axial resolution of A-mode scans using system identification. Long-term plans call for extending the system identification concept to B-scans and phased array technology.
