Research On Ultrasonic Speckle Motion Tracking And Elastic Imaging Based On Bayesian Theor

Ultrasonic speckle motion tracking is a crucial step in ultrasonic elastography.Many scholars are devoted to the research of ultrasonic motion displacement tracking.Ultrasound elastography is a new imaging method that has been gradually applied in the past 20 years.It is of great significance for the diagnosis of cancer diseases such as cancer.For example,in the diagnosis of breast tumors,ultrasound elastography has been successfully applied to the noninvasive differentiation of breast lesions.The specific work in this thesis is divided into two parts:two-dimensional and three-dimensional ultrasonic elastography.In two-dimensional ultrasound elastography,a region growing Bayesian algorithm with local displacement fitting is proposed.The basic idea of the algorithm is to replace the traditional cross-correlation value with the maximum posterior probability value of Bayesian inference and local displacement,change the direction of activity growth,and improve the quality of strain images.In 3D ultrasound elastography,a 3D ultrasound elastography based on Bayesian theory is proposed.The two-dimensional ultrasonic strain image and three-dimensional ultrasonic strain image can better show the displacement of tissues and organs.Performing three-dimensional speckle motion tracking can reduce motion tracking caused by more than plane motion.Obtaining deformed three-dimensional strain data provides clinicians with more The plane or vertical plane assists diagnostic information.The 3D ultrasonic elastography algorithm based on Bayesian theory is based on the 3D block matching algorithm.This method extends the 1D Bayesian estimation method to the 3D ultrasonic elastography of mesoscale prediction of 3D ultrasonic strain images.The idea is to fuse neighborhood information to optimize displacement.2.A three-dimensional ultrasonic elastography based on Bayesian theory is proposed.The three-dimensional ultrasonic elastography solves the problem of signal decorrelation in two-dimensional ultrasonic elastography.This method extends the one-dimensional Bayesian estimation method to three-dimensional ultrasonic strain images to obtain high-quality three-dimensional ultrasonic elastography.The three-dimensional ultrasonic elastography method based on Bayesian theory is based on three-dimensional block matching.The calculation method is to introduce the similarity matrix corresponding to each estimated point calculated by the block matching method and the motion of adjacent estimated points into the Yes method to optimize displacement.The experimental results of simulation data,phantom data,and real data show that the 3D ultrasonic elastography method based on the Bayesian method has a higher contrast-to-noise ratio than the cross-correlation method based on the table.Bayesian method can remove noise points in displacement estimation,remove abnormal strain values in strain images,and obtain high-quality ultrasonic strain images.In the experimental part,the computer simulation data,phantom phantom data and a group of pathologically confirmed ultrasound data of breast cancer patients were used to compare the improved algorithm proposed in this thesis with the traditional regional growth motion tracking algorithm.The experimental results show that the addition of Bayesian theory and the introduction of local displacement fitting are useful for improving the accuracy of motion tracking.The displacement error obtained by the improved algorithm has small tracking error,low mean absolute error(MAE)value in computer simulation data,and high motion compensation cross-correlation(MCCC)value in phantom phantom data,in real ultrasound data The remodeled volume images obtained are more clearly visible for breast tumors.At the same time,the strain image obtained by the improved algorithm has a higher contrast to noise ratio(CNR)value.The three-dimensional ultrasound data uses two sets of phantom phantom data to compare the Bayesian algorithm and the traditional cross-correlation algorithm.The experimental results show that the improved algorithm has a higher contrast to noise ratio and lower motion tracking error.It shows that the Bayesian method can effectively remove the noise points in the displacement estimation,reduce the abnormal strain value in the strain image and finally obtain the high-quality ultrasonic strain image.