The Research On The Reconcstruction Algorithm Of Transmission-type Ultrasound Tomography Imaging

Relative to X-CT, Transmission-Type Ultrasound Tomography has thecharacteristics of lower cost and slighter radiation, because of these, it is widely usedin breast cancer diagnosis and industrial NDT. However, this technology is still in thetheoretical stage and research has not yet reached a very high level. Reconstructionalgorithm is the most central place of the technology, so this article focuses oncomputer implementation of reconstruction algorithm and how to solve frequentlyasked questions and improve the algorithm.Firstly, this article studies the filtered back projection reconstructionalgorithm.Additionally, this article simulates the algorithm in three ways--parallelbeam mode, ring matrix mode and fan beam mode. Besides，this article studies theproblems in computer implement including selecting the interpolation function,selecting and designing the filter and how to calculate the beam.Algebraicreconstruction algorithm is brief and clear, which converts the reconstruction intosolving equations.However, in the computer-implemented process, becauseprojection coefficient matrix is often very huge, so dealing with it will occupy a lotof memory and time. To solve these problems, this article proposes an algorithm forfast solving projection coefficient matrix and accelerating the iterative convergence.Secondly,because ultrasonic propagation path is not a straight line，so, theimage which we solve out assuming that ultrasonic propagation path is a straight linewill inevitably have error. In previous research work, the scholars did not directlyaddress this issue, they only studied how to reduce the error. To solve this problem,this article introduces in ray tracing method commonly used in geologicalexploration to track the path of the ultrasonic propagation. According to this path, wemake some improvements to the traditional algorithmas.Finally,this article studied image fusion based on wavelet transform.There aresix kinds of fusion rules we used in this article and we gived a new rule based on theaverage gradient and variance,and this rule has achieved good results.