Method Of Non-invasive Temperature Estimation Based On Spectral Image

High intensity focused ultrasound(HIFU)technology is one of the most ef-fective and potential treatment methods about tumors in the 21st century,which has been widely used in clinical medicine.Tumor treatment technology with HIFU mainly takes advantage of HIFU's thermal effect to destroy and ablate the tumor tissue.In order to protect normal tissue,it is especially crucial to monitor the real-time temperature during the HIFU treatment.In this paper,B-mode ultrasonic images of the fresh pork tissue are acquired before and after irradiating by using HIFU experimental system of PRO.Then,the previous relevant investigation can be carried out smoothly,such as matching the B-mode ultrasonic images with the real-time temperature and pre-processing the experimental data.The main work is shown as follows:(1)Pre-processing the B-mode ultrasonic images of the fresh pork tissue,which mainly includes the selection and match of B-mode ultrasonic image before and after irradiation,converting the treatment image into the gray-scale image and image cropping,etc.Then the follow-up research was carried out.(2)Image enhancement and registration.Firstly,the B-mode ultrasonic im-age's pseudo-color is enhanced by using pseudo-color enhancement algorithm of spatial domain and gray-level transform pseudo-color enhancement algorithm,which makes the color information of the B-mode ultrasonic image more obvi-ous.Secondly,the SIFT registration method and the SURF registration method is introduced.The CSURF registration method,which takes full advantage of the color information and global information of the B-mode ultrasonic image.Using CSURF registration method to register the image after pseudo-color enhancement,the treatment area is obtained accurately.After this step,these image data is pro-vided for the following study about non-invasive temperature estimation.(3)A method of non-invasive temperature estimation for HIFU therapy based on gray level co-occurrence matrix of Hadamard spectral image(Hadamard-GLCM)is presented.Firstly,irradiating the fresh pork in vitro by HIFU,the subtraction image is obtained from real-time B-mode ultrasound images before and after irra-diation.Then,the spectral image by Hadamard transform from the subtraction image can be acquired.Finally,the Hadamard-GLCM's inertia moment of spectral image can be used as the temperature information parameters.The experiments showed that the Hadamard-GLCM's inertia moment of single group data can be linear fitted well with the temperature,whose fitting degree close to 1 with low er-ror,high resolution and strong fault tolerance,but also the multiple sets of data's Hadamard-GLCM inertia moment do.Compared with the traditional tempera-ture estimation method,this method has obviously advantages that it is able to provide the real-time information for the noninvasive temperature estimation of HIFU treatment.