Study On The Registration Method Of B-Mode Ultrasonic Monitoring Images Before And After HIFU Treatment

High Intensity Focused Ultrasound(HIFU) technology has been widely used in clinical medicine becauseit is non-invasive and without radiationor side effects. In recent years, HIFU technology has achieved remarkable results in the resection of prostate cancer, uterine fibroids, metastatic liver tumors, pancreatic cancer, the treatment of kidney cancer and breast cancer treatment.Through a large number of research and visitall over the years, the use of HIFU for the purpose of medical treatment has been accepted by everyone.As ultrasound imaging has the advantage of easier integration and real-time imaging, B-mode imaging is typically usedin HIFU systems. The monitoring image before and after the treatment weresubtracted to obtain the image of the treated area and determine the therapeutic effect. Since the patient’s position may bechanged after the treatmentresulting in that the images acquired areshifted, the registration of ultrasonic monitoring image before and after HIFU treatment is necessary.In this paper, we focus our effects on the study of registration methods ofB-mode ultrasonic monitoring images before and after treatment combiningitsintrinsic property. Details are as follows:Firstly, the related theories of high-intensity focused ultrasound technologyincluding the basic principles, background and status of focused ultrasound therapy guiding and monitoring methods are introduced in detail. Then the status of ultrasound treatment efficacy evaluation are described and registration problemsbefore and after HIFU treatment B super monitor image are proposed, highlighting the significance of this subject.Secondly, the basic concepts and common registration methods of image registration are described. By introducing the way and characteristics of ultrasound imagingand analyzing the problems and difficulties in ultrasound image registration, the best method for ultrasound image registration is proposed.The basic concepts of medical image and several commonly used methods of image registration are explained and a detailed comparison of the advantages and disadvantages of several algorithms has been done. Combined with ultrasound imaging features, the key problems of the ultrasound image registration areanalyzed.Thirdly,the combination of image pre-processing and Fisher-Tippett distribution of feature based image registration method is proposed. The characteristics of speckle noisein ultrasound imagesareanalyzedat first, then several common image preprocessing methods arecompared.A median filter is used for ultrasound image pre-processingand image features are extracted through thedistribution made by Fisher-Tippett.It not only can eliminate loose spot noise, but also retain good image edge information which can be used for registration algorithm based on thefeatures.Fourthly, animproved SURF registration algorithmis proposed. In conventional SURF algorithm,the number of matching feature points are less registration and the accuracy is lowdue to the influence of speckle noise.First of all, feature points are extracted from the monitoring image before and after the treatment with classical SURF feature detector. Then we use the Euclidean distance to match the feature vectorsand eliminate erroneous matching points with the modifiedrandom sample consensus(RANSAC) algorithm. To optimize registration parameters, we make full use of the inherent properties of speckle features distribution of the ultrasound image. Finally, the relationship between the images is mappedto complete registration with the least squares method. Experimental results show that the algorithm can automatically search corresponding feature points from the ultrasonic monitor image before and after treatment which is more efficient and more accurate than the standard algorithms.