Feature Extraction Of Ultrasonic Signal And Identification Of Denatured Biological Tissue During HIFU Treatment

High-intensity focused ultrasound(HIFU)therapy is a non-invasive tumor treatment modality for conducting high-temperature thermal therapy.HIFU has the capability to instantly generate high temperature in the treatment region by delivering acoustic energy at a distance from the source,and to kill the cancer cells in the treatment region without damaging normal tissues outside the treatment region.It can reflect the effect of HIFU treatment by monitoring the temperature and damage of the biological tissue treatment region,which is of great significance to ensure the safety and efficiency of HIFU treatment.This paper focuses on four aspects of the monitoring of denatured biological tissues in HIFU treatment based on the characteristics of ultrasonic signals:(1)Interception of the ultrasonic scattered echo signal in HIFU irradiated region and denoising of the ultrasonic signal.(2)Studying the correlations between various characteristic parameters of ultrasonic signals and temperature.(3)Based on the phase space theory,studying the nonlinear characteristics of ultrasonic scattered echo signals.(4)Studying the identification methods of denatured biological tissue based on clustering analysis.The specific work is as follows:(1)A signal interception method based on generalized Stransformation and time-frequency entropy was proposed for intercepting ultrasonic scattered echo signals in the HIFU irradiation region.The boundary between the normal region and the irradiation region was determined by the peak of the time-frequency entropy of the ultrasonic echo signal based on generalized S-transform.The results show this method can improve the reliability of intercepting signals.In view of the shortcomings of the traditional empirical modal decomposition(EMD)denoising algorithm,a denoising method of the ultrasonic signal based on DFA-VMD was proposed by combining the detrended fluctuation analysis(DFA)algorithm and variational modal decomposition(VMD)algorithm.This method not only effectively solves the mode aliasing phenomenon in the EMD process,but also overcomes the shortcoming of pre-setting the number of modes in the process of VMD.(2)The correlations between the characteristic parameters of ultrasonic signals and the temperature of biological tissues were studied,including the ultrasonic velocity,ultrasonic attenuation coefficient,and ultrasonic frequency shift characteristic parameters.In the study of ultrasonic attenuation coefficients,a new attenuation coefficient measurement method based on the characteristics of the quality factor(Q)value of ultrasonic echo signals was proposed to overcome the shortcomings of the traditional time and frequency domain acoustic attenuation coefficient measurement methods.In the study of the ultrasonic frequency shift,the maximum frequency of the peak of the auto-regressive spectrum(AR)of the ultrasonic echo signal was used to characterize the center frequency of the ultrasonic echo signal.The results show that when the temperature is between 37?to 53?,the ultrasonic velocity of biological tissues rises with increasing temperature.When the temperature is between 53?to 63?,the ultrasonic velocity rises slowly.When the temperature reaches 63?,the ultrasonic velocity of biological tissues reaches a maximum,and the ultrasonic velocity hardly changes after 63?.When the tissue temperature is between 37?to 63?,the ultrasonic attenuation coefficient increases with increasing temperature,the center frequency of the ultrasonic scattered echo signal decreases with increasing temperature.After 63?,the ultrasonic attenuation coefficientgradually slows down until it remains unchanged,and the center frequency of the ultrasonic scattered echo signal does not change with increasing temperature.At the same time,The correlation coefficient between the attenuation coefficient and temperature measured by the Q value method reaches 0.9796,and the standard deviation is small.Compared with other methods,The Q value method can obtain more accurate and stable ultrasonic attenuation coefficient measurement results.(3)Based on the phase space reconstruction theory,the nonlinear characteristics of ultrasonic scattered echo signals from biological tissues were studied.The reconstructed three-dimensional phase space trajectory shows that when the biological tissue is denatured,the phase space trajectory of the ultrasonic scattered echo signal is divergent and chaotic.The phase space trajectory of the ultrasonic scattered echo signal of normal tissue converges well.In view of the shortcomings of multi-scale permutation entropy(MPE)and multi-scale weighted permutation entropy(MWPE),the refined composite algorithm was introduced to improve MWPE and the refined composite multi-scale weighted permutation entropy(RCMWPE)was proposed.The results show that RCMWPE not only measures the complexity of signal including amplitude information,but also improves the stability and reliability of multi-scale entropy.The intra-class distance of RCMWPE is less than MPE and MWPE,and the inter-class distance of RCMWPE is greater than MPE and MWPE.(4)The identification methods of denatured biological tissue based on clustering analysis were studied.The genetic algorithm(GA)optimized support vector machine(GA-SVM)and GK fuzzy clustering methods were used to distinguish non-denatured and denatured tissue according tothree characteristic parameters of the ultrasonic scattered echo signal,including multi-scale permutation entropy,multi-scale weighted permutation entropy,and refined composite multi-scale weighted permutation entropy.The results show that under the experimental conditions of this paper,the clustering effect and recognition ability of the clustering method based on the refined composite multi-scale weighted permutation entropy feature are better than that based on multi-scale permutation entropy and multi-scale weighted permutation entropy.Compared with GA-SVM,The clustering methods based on the GK fuzzy clustering have higher recognition rate and shorter running time.This proves that the GK fuzzy clustering method can better solve the problem of fuzzy denatured characteristics of biological tissues.The identification method of denatured biological tissue based on RCMWPE-GK fuzzy clustering can better identify non-denatured tissues and denatured tissues,and the recognition rate is as high as 95.5%.