Deconvolution Study Of Ultrasonic Medical Image Based On Higher Order Spectral And Wavelet Theory

Medical ultrasound imagings use the detected variances of the sound parameters, such as reflectivity, to display different tissues / organs of human body and the inner structures. It has been widely used in medicine and in clinics, due to its low-cost, no-trauma and safety. But low resolution is the shortcoming of the ultrasound images. It impacts the diagnosis. Therefore improving resolution of ultrasound images is an important and attractive research area.The reason of the low resolution in ultrasound images is intrinsically due to the convolution between the reflectivity of detected objects and the impulse response of the system, which produces images. The impulse response and the convolution are the key factors. Therefore, if there have some fine methods to find out the impulse response of a specific system and to deconvolve to remove the two blurring factors, the resolution of ultrasound images will be improved.In this thesis the higher-order spectrum analysis and wavelet theory were used to deconvolve 121 ultrasound images including abdomen, thyroid gland and breast. The higher-order spectrum analysis was used at first to estimate impulse response, and then the wavelets was used to denoise and to deconvolve in wavelet domain. The results show that the inner hierarchies and textures are clearer and finer even more.