The Research On Ultrasonic Elastography Algorithm Based On Keypoints Tracking Strategy

The elasticity information of biological tissues has very important reference values to some disease diagnosis(e.g.: breast cancer,thyroid nodule,liver tumor,prostatic cancer,etc),since changes in tissue's elasticity and rigidity are usually related to the pathological changes.Recently,the manual palpation is still a common diagnostic method due to its simplicity and easy implementation.However,results of this method are prone to be affected by the subjective judgments of physicians.On the other hand,the traditional ultrasound diagnostic equipment can hardly detect lesions whose ultrasonic echo characteristics are similar to the normal tissue.Therefore,the elastography technique developed rapidly in recent years.The elastogram obtained from ultrasonic elastography can reflect the information of the elasticity distribution within tissue directly,which can provide a new imaging method in clinical diagnosis for these diseases.In this thesis,the one dimensional static imaging methods in the ultrasonic elastography were studied.The work and the main contributions of this thesis are as follows:First,a new fast peak-searching algorithm(FPSA)was proposed with an optimization strategy in the process of displacement calculation.This method locates the key points in the ultrasonic signal directly based on the monotonicity of the function and without needing the partition of RF signals to different segments by slipping a window and also without needing derivative operations.As a result,the processing time as well as the operation error can be significantly reduced.Second,in order to further reduce the computational requirements,a single peak-searching(SPS)algorithm was proposed.This method only searches the maximum(or minimum)values in the ultrasonic RF signals as key points for motion tracking in tissues.As compared with the FPSA,the imaging speed can be further accelerated without affecting the imaging quality since this method is interested only in one kind of extrema and hence one half of computational requirements can be reduced.In this thesis,the ANSYS(ANSYS Inc.,Canonsburg,PA,USA)was used to build a FEM phantom as a simple model of virtual human tissue while the Field ? is used to generate the simulated ultrasonic RF signals,based on which simulated experiments were implemented.Simulation results show both qualitatively and quantitatively that these two algorithms can be applied in a larger dynamic range and can be implemented much faster as compared with other methods such as the time-domain cross correlation(CC)based methods.