Deconvolution Study Of Ultrasonic Medical Signals

Medical ultrasound images has been widely used in medicine and in clinics, due to its imaging immediately, no-trauma and low price. But low resolution is the shortcoming of the ultrasound images. The reason of the low resolution in ultrasound images is primely due to the convolution between the reflection of the detected tissue and the impulse response(called'System Function'or'Point Spread Function') of the ultrasound system, also the noise of the environment is added, which result in the images blurred. If we denoise the images, find out the impulse response of the system, and then deconvolution the images with the impulse response , we can recover the original images of the detected tissue. It is clearly that we will receive more clear images then the detected ones.Wavelet is a new branch of mathematics, which was greatly developed in resent years. It is a new effective time-frequency domain analysis method after Fourier Transform, it can analyzes the signal at different frequencies with different resolutions, which called the multiresolution analysis (MRA). So it is suitable to analysis the signal with sharp information, such as boundary and edges.The higher order spectral method is based on that the cumulants of Gaussian signals, which is higher than two order, are zero. The separation of the Non_Gaussian signals (the system impulse response h(n)) and Gaussian signals (tissue signal x(n) and noise) can be straightly forwarded in their higher order spectral domain.This thesis was based on the above two method: got A-line from Belson 200A B-mode ultrasonic instrument, imaging them with deconvolution method. There were three main task in it: first,find out system function with higher order spectrum method; second, denoise it with wavelets; and last, accomplish deconvolution in wavelets domain. The result showed that the resolving power was greatly improved.