|M.Sc Student||Hevel Guy|
|Subject||Improving Resolution of Ultrasound Images Using Optimization|
|Department||Department of Biomedical Engineering||Supervisor||Professor Emeritus Dan Adam|
Ultrasound pulse-echo imaging is a modality based on interrogation of tissue by high frequency focused acoustic wave of short duration. Due to internal impedance changes, part of the incident energy is reflected and can be measured to obtain information about the internal structure of the object. An image is formed by geometrically aligning echo signals, but it contains also “speckle noise” due to reflections of the interrogating pulse from a large number of small scatterers spaced densely within the tissue. Several factors have been identified in the literature as those contributing to the limited diagnostic utilization of ultrasound images. Among the most basic are pulse length, beam width and variation in echo strength. The finite signal bandwidth of the ultrasonic transducer is the major reason for the low axial resolution, whereas the non-negligible beam width highly contributes to the low lateral resolution. Under simplifying assumptions, the recorded ultrasonic image can be modeled as a convolution between a pulse shape and the reflectivity of the medium with an additive noise. The convolution with the pulse "smears" out fine details in the reflectivity and makes interpretation of closely spaced reflectors difficult. Major research effort has been invested to remove the degradation in resolution caused by the transducer, so that the remaining signal characterizes the true tissue properties, and by doing so the diagnostic quality of the ultrasonic images can be improved. The purpose of this research is to develop an iterative algorithm to improve the resolution of ultrasound images, based on optimization methods. The algorithm is designed to use an a-priori information on the PSF function and the sparse nature of the image. When added to the deconvolution function, sparseness can be a powerful constraint that regularized the ill-posed problem. Minimization of the objective function leads to significant resolution improvement of strong reflectors and noise reduction in two-dimensional ultrasound images.