Study On New Methods For Assessing Liver Fibrosis Using Ultrasound Backscattering Homodyned K Imaging

China is a big country with liver diseases,with about 300 million people suffering from different types of liver diseases.Liver fibrosis is the common pathological change of chronic liver disease,so it is very necessary to prevent liver disease early through the early diagnosis of liver fibrosis.Ultrasonic imaging has been widely used in the detection of liver fibrosis due to its low cost,good real-time performance and no ionizing radiation.And in many ultrasonic detection technologies,using mathematical statistical distribution to describe and model the scattering statistics,so as to carry out ultrasonic backscattering statistical parameter imaging is a hot trend.Based on the conventional ultrasound equipment platform,this paper implements a new method for evaluating liver fibrosis based on ultrasonic backscatter homodyne K imaging,and adopts Empirical Mode Decomposition（Empirical Mode Decomposition,EMD）and Noise Empirical Mode Decomposition（Noise-modulated Empirical Mode Decomposition,NEMD）.The two signal decomposition technologies solve the problem that the target ultrasonic backscatter signal（liver fibrosis signal）is affected by the non-target backscatter signal（such as liver parenchymal signal,fat signal,etc）.The main research contents of this paper are as follows:1)Study on the Evaluation of Hepatic Fibrosis by Ultrasonic Backscatter and Homodyne K Imaging:A new method for liver fibrosis evaluation based on Homodyned K distribution model based on RSK and XU parameter estimation was proposed.Based on the data of 94 patients with hepatic fibrosis without steatosis（F0=14,F1=15,F2=27,F3=18,F4=20）,the parameters of k _RSK,μ_RSK,k _XU andμ_XUwere calculated by RSK and XU parameter estimation method based on sliding window technique,respectively.The parameter matrix and parameter image were constructed,and the calculation results were statistically analyzed.The results showed that the parameters k _RSK and k _XU reached the optimal evaluation results when diagnosing liver fibrosis≥F1（that is,with or without liver fibrosis）,and the area under the ROC curve reached 0.70 and 0.75,respectively.Combining the k _RSK and k _XU parameter images and box plots,it can also be found that the estimated k _RSK and k _XU parameter values show a corresponding change trend with the liver fibrosis grading.Therefore,the ultrasound backscatter homodyne K imaging proposed in this paper can initially determine the presence or absence of liver fibrosis,and the XU method is better than RSK in parameter estimation.2)Study on the evaluation of liver fibrosis by ultrasonic homodyne K imaging based on empirical mode decomposition:A new method for the evaluation of liver fibrosis by EMD Homodyned K-parameter imaging was proposed to deal with the influence of noisy backscattered signals such as liver parenchyma and fat on the statistical evaluation of liver fibrosis signals.The impact of statistical evaluation has improved the accuracy of liver fibrosis detection.The data of 94 cases of liver fibrosis without steatosis（F0=14,F1=15,F2=27,F3=18,F4=20）and 143 cases of liver fibrosis with fatty liver（F0=13,F1=34,F2=18,F3=28）were used,respectively.F4=41),and the EMD was decomposed to obtain the first Intrinsic Mode Function（IMF₁）and the second Intrinsic Mode Function（IMF₂）,which contain the most information.IMF₁ and IMF₂ is used to calculate the k _IMF1,μ_IMF1,k _IMF2,andμ_IMF2 parameter values through the sliding window technology using the XU parameter estimation method,the parameter matrix and parameter image were constructed,and the parameter calculation results were statistically analyzed.The results show that when evaluating liver fibrosis data without steatosis,compared with traditional homodyne K imaging,the parameterμ_IMF1 improves significantly when≥F1 classification,and the area under the ROC was increased from 0.59 to 0.77,accuracy,Sensitivity and Specificity have been increased to 75.53%,75.01%and78.57%respectively.The imaging result of parameterμ_IMF1 was improved compared with traditional homodyne K imaging.It can be found that the image brightness in the ROI has a certain change rule with liver fibrosis classification.The box plot of parameterμ_IMF1also shows that the parameterμ_IMF1 has a direct relationship with liver fibrosis classification.When evaluating liver fibrosis data without steatosis,the results showed that EMD was ineffective in dealing with fatty signals mixed with liver fibrosis signals.Studies have shown that in the evaluation of liver fibrosis without steatosis,EMD preliminarily eliminates the influence of backscattered signals from non-target tissues such as liver parenchyma on liver fibrosis signals,and improves the performance of ultrasonic homodyne K imaging in detecting liver fibrosis.The best performance is when judging the presence or absence of liver fibrosis.However,the empirical mode decomposition method cannot effectively solve the problem of liver fibrosis fatty liver coexist.3)Research on the evaluation of liver fibrosis by ultrasonic homodyne K imaging based on empirical mode decomposition of noise:A new method based on NEMD ultrasonic Homodyned K imaging was proposed to evaluate liver fibrosis,in order to solve the influence of higher echo fat signal on liver fibrosis signal,and improve the accuracy of traditional Homodyned K imaging in liver fibrosis evaluation when fat coexists.Using the data of 143 cases of liver fibrosis coexist with fatty liver（F0=13,F1=34,F2=18,F3=28,F4=41）,NEMD and Noise-assisted Correlation Algorithm（NCA）were used to decompose the IMF₁ and the IMF₂,and then using sliding window technology,the XU parameter estimation method is used to calculate the parameter values of k _IMF1,μ_IMF1,k _IMF2,andμ_IMF2,the parameter matrix and parameter image were constructed,and the parameter calculation results were statistically analyzed.The results show that compared with the traditional homodyne K parameterμ,the parameterμ_IMF2 has improved in≥F1 classification and≥F2classification,especially in≥F1 classification,the effect is the best（area under the ROC value reaches 0.81,Accuracy reaches 79.72%,Sensitivity reaches 80.23%,Specificity reaches 76.61%）.The imaging result of parameterμ_IMF2 shows that the image brightness in the ROI shows a certain degree of regular changes with the liver fibrosis grade,and the parameter box plot shows the inevitable relationship between the parameterμ_IMF2and the liver fibrosis grade.Studies have shown that homodyne K imaging based on NEMD improves the performance of liver fibrosis detection when fatty liver coexists,especially the early detection of liver fibrosis.