The Estimation Method Of The Ablation Zone Of Liver In Radiofrequency Ablation Based On The Ultrasound Imaging Models

Radiofrequency ablation(RFA),which is minimally invasive and effective method for tumor therapy,has received more and more attention and is promising to be the replacement of clinical surgical treatment.Ultrasound imaging,as an auxiliary monitoring device for RFA surgery,which is the most widely used method because of its character of real-time and low cost.A limitation of conventional B-mode imaging is the inability to visualize tissue scatterer properties.To accurately estimate ablation zones,Nakagami-m imaging is used to efficiently describe the statistical properties of the internal scattering of liver tissue.However,conventional Nakagami-m imaging failed to weaken the effect of acoustic posterior shadow in B-mode image.In this thesis,3 improved algorithms are proposed based on the Ultrasound imaging features and the physical processes of RFA.Firstly,in order to weaken the acoustic posterior shadow effect,the window-modulated compounding(WMC)Nakagami-m algorithm was proposed.The proposed method based on the synthesis of Nakagami images by using windows with different side lengths is able to position ablation zones accurately.Then,a fast Nakagami-m imaging algorithm was used for computational acceleration.Instead of estimating the Nakagami parameter pixel-for-pixel,an average filter is used to do convolution calculation with the envelope image.Furthermore,the thesis proposed a new strategy for the postprocessing of Nakagami images by Gaussian approximation as an alternative of polynomial approximation.The experimental results show the acoustic posterior shadow effect is depressed effectively by the WMC Nakagami-m algorithm and the operation speed of Nakagami-m imaging algorithm is improved combined with the fast Nakagami-m imaging algorithm.The estimations of ablation zone with Gaussian approximation are more accurate than that with polynomial approximation.The relative coefficient between estimations of Gaussian approximation and artificial measured value is 0.97,which is higher than that with polynomial approximation.