Study On The Recognition Of Denatured Biological Tissue During HIFU Therapy Based On Image Processing

A new technique called High Intensity Focused Ultrasound(HIFU)is used to treat benign and malignant tumors,which concentrates ultrasound on local tumor areas outside the human body.Due to the penetrability and high intensity of ultrasound,ultrasound penetrates directly to the treatment area,making the target area reach over 65? in a short time.Consequently,irreversible coagulation necrosis occurs in the tumor tissue.However,the surrounding normal tissue is unaffected.The technique is non-invasive in tumor treatment.During HIFU therapy,it is important to monitor and detect the damage of biological tissue in real time.In this paper,fresh pork in vitro was adopt as the experimental object.We obtained ultrasonic images before and after HIFU treatment by B-ultrasound.Then,we extracted the features from processed ultrasonic images.Finally,we detected tissue lesion according to the characteristics of the interested region.The main work and innovations are as follows:Firstly,we elaborated the therapeutic principle of HIFU technology,the research background,the methods of identifying HIFU tissue degeneration.Then we introduced the experimental system and the method.We preprocessed the ultrasonic images,including filtering,graying and difference processing.All of procedures were prepared for the identification of HIFU region denaturation.Secondly,a method of identifying denatured biological tissue based on statistical characteristics of ultrasound images was presented.Seven statistical features of grayscale image were extracted from ultrasonic images.The inter-distance with support vector machine(SVM)wasemployed to evaluate the ability of several statistical features to recognize the denatured biological tissues.Finally,with the method of OTSU to traverse different threshold values,we could obtain the corresponding inter-class variance without setting the parameter threshold artificially.The experimental results showed that the statistical characteristics could effectively distinguish the denatured tissue from the undenatured tissue.After SVM training,we could obtain the classification threshold with the method of OTSU from the sample of features.The threshold could recognize denatured biological tissue.Thirdly,a method of identifying denatured biological tissue based on PCA was proposed.The grayscale run matrix and eleven texture features were extracted from the ultrasonic image in four different directions.PCA was used to reduce the dimension of multidimensional features in four directions.According to the cumulative contribution rate,we could get the principal component data after dimension reduction.The principal component was input into SVM for classification and recognition in four directions.The experimental results showed that the multidimensional texture features of the gray range-running matrix with PCA could effectively identify denatured biological tissues.In this paper,we researched on the methods of detecting tissue denaturation by processing ultrasonic images,combining with the OTSU,PCA and machine learning,which were meaningful for improving efficacy during HIFU treatment.