Research On Displacement Denoising Algorithm For High-precision Ultrasonic Elastography

Medical ultrasound elastography technology can provide medical staff with elasticity related data of human local soft tissue,so as to reflect the degree of tissue sttifness,which gives an important basis for disease detection and diagnosis.The appearance of ultrasound elastography has made up for the shortcomings of doctors’ palpation diagnosis and provided a new reference for disease diagnosis.Ultrasound elastography has become a hot topic in the field of medical imaging.However,due to the limitations of the ultrasound equipment itself,in addition to the impurities of radio frequency signal or processing algorithms in the subsequent imaging operations,the ultrasound displacement image may contain noise.The presence of noise in the displacement image has an adverse effect on the subsequent strain imaging and elastic modulus distribution analysis.Therefore,it is of great significance to study the denoising of displacement images in ultrasound elastography.Partial differential equations technologies are important methods in machine vision and digital image processing with a solid mathematical theory basis,which can provide multiple numerical solutions for regularization equations.The first part of this paper uses partial differential equations to denoise local displacement images.The second part introduces the inverse gradient problem framework,so that transforms the displacement image denoising process to iteratively solving the pressure Poisson equation.Firstly,based on the assumption of incompressibility of biological tissues,a second-order Laplacian smoothing operator is introduced into the classical PDE algorithm,an improved regularized energy functional model is proposed,and the Poisson’s equation corresponding to the model is derived by using the variational method.Finally iteratively finds the optimal solution of the energy function through the finite difference method;Secondly,different from traditional two-dimensional strain image,instead of considering the gradient of two directions,added the axial displacement in the y direction and the gradient of the lateral displacement in the x direction in strain imaging.The shear strain in four directions is integrated,and the inverse gradient problem frame is used to find optimal displacement vector.The inverse gradient optimization problem is converted into a solution to the pressure Poisson equation,which reduces the difficulty of calculation and improves the calculation speed.Finally,The computer simulation ultrasound data,tissue-mimicking phantom data and real breast tissue data were used to verify above two methods.The experimental results show that the two denoising algorithms proposed in this paper can suppress the noise in the displacement image and improve the image accuracy.