Study On Monitoring Methods Of HIFU Treatment Based On Ultrasound Images

High-intensity focused ultrasound(HIFU)therapy is a new elinical d iagnosis and treatment technology.It is widely used in clinical treatment because of its advantages of non-invasive treatment,minimal trauma to th e patient,and high safety and efficiency.HIFU treatment is ultrasound pr oduce high temperature instantaneously and the high temperature could c ause the cell protein denature and necrosis in the body.It is necessary to e ffectively monitor the HIFU treatment.Ultrasound image monitoring tech no logy has become a important method for high-intensity focused ultraso und monitoring because of its advantages such as less harm to patients,lo w technical cost,and compatibility with hyperthernia instruments.In this paper,a high-intensity focused ultrasound transducer was used to irradiat e the fresh isolated pork tissue.Ultrasound images before and after irradia tion were acquired in realtime through B-mode images.Ultrasound imag es were used to monitor the high-intensity focused ultrasound treatment p rocess and conduct research.The main tasks as follows:Firstly.Introduced the principle of HIFU technology and its development at home and abroad,and focused on the research of ultrasound image processing methods for HIFU treatment at home and abroad,which further highlighted the research significance of this topic.Secondly.This paper introduces the design scheme of the experimental system and the ultrasonic image acquisition process in detail.The pretreatment scheme of ultrasound images is introduced.It mainly deals with image gray-scale processing,wavelet filtering,differential image processing of ultrasound images,etc.It is prepared for the extraction of texture features of ultrasound images furtherThirdly.Development of an ultrasonic noninvasive temperature estimation.Extracting the characteristic values of gray mean values,fractal box dimensions,normalized gradient co-occurrence matrix second-order parameter gradient averages,mixed entropy,and deficit moments of the ultrasound image after differential processing,using the least square method.The fitting principle linearly fits the characteristic parameters and temperature,and adds the average value of the linear correlation coefficient as the evaluation criterion of the fitting degree.The experimental results show that there is a linear correlation between gray mean value,box dimension,gradient average,hybrid entropy,deficit moment and temperature.Among them,the gray mean value and the fractal box dimension can only be at low temperature.The low temperature stage reflects the information about temperature well.With the increase of temperature,the tissue denaturation becomes worse and the temperature can no longer be effectively reflected.The mixture entropy extracted by the Gabor-GGCM method can accurately estimate the changes of temperature information in the low temperature phase,and the gradient average can manifest in the high temperature segment,the information of temperature change can be better reflected;and the average values of the linear correlation coefficient of the three second-order parameters about temperature are extracted by the Gabor-GGCM method are 0.9804,0.9854,and-0.9644,respectively.Higher than the gray average method and the fractal box dimension.The noninvasive temperature estimation method based on Gabor transform and gray gradient co-occurrence rmatrix proposed in this paper provides a new way for ultrasonic noninvasive temperature measurement.Fourthly.Conduct research on tissue damage assessment.The characteristic parameters such as gray mean value,information entropy,conditional entropy,and linear correlation coefficient of ultrasonic images were extracted,and the tissue damage was evaluated by support vector machine.Experimental results show that the four characteristic parameters of gray mean,information entropy,conditional entropy,and correlation coefficient all can identify whether the tissue is damaged and have different threshold intervals.After full training,these four parameters can be used for the division of tissue damage thresholds.Among them,the average recognition value of two characteristic parameters of gray mean value and conditional entropy is lower than that of information entropy and linear correlation coefficient;conditional entropy,information entropy,and linear correlation coefficient better on the identification of undamaged tissue than gray Mean value.Gray mean value,information entropy,and linear correlation coefficient are better than conditional entropy in identifying damage tissue.In this paper,based on the ultrasound images,some researches have been carried out from the perspectives of ultrasonic noninvasive temperature estimation and tissue damage assessment,which proposes new ideas for HIFU treatment methods in monitoring process.