Research On Biomedicine Image Denoising Based On Partial Differential Equation

In medical imaging, type-B ultrasonic imaging with the advantages of low cost, lossless detection, real-time imaging and so on, is widely applied to medical diagnosis. The origin of type-B ultrasonic leads to the unavoidable speckle noise. To solve the problem that when denoising the type-B ultrasonic the edge information is nearly missed and thus the sequential process is more complicated, a revised denoising algorithm particular for type-B ultrasonic is proposed here.Firstly, the advantages and disadvantages of partial differential equation in the denoising of images during its pre-processing stage is discussed. ROF partial differential equation denoising model retains clear boundaries but results in stage effect. Y-K partial differential equation denoising model suppresses the stage effect but retains the boundaries weakly. By synthesing the models above,a new comprehensive image RYK denoising model is proposed.After analyzing the origin of type-B ultrasonic image multiplicative speckle noise,a denoising and edge-retaining term is conducted and denoising model based on type-B is proposed.Secondly,based on research into general P-M partial differential denoising model expanding coefficient, a revised methodology for calculating expanding coefficient is proposed. Utilizing adaptively chosen expanding threshold K, the new coefficient chooses different expanding coefficients in texture sections and the other sections,thus keeps the texture and edge information. This new methodology for calculating the coefficient is then applied to RYK model specially revised for type-B ultrasonic. The experiment shows the model can process speckle noise and other multiplicative noise as well and further enhance the denoising and edge-keep capabilities for type-B ultrasonic image.