Study On CT Image Recognition Of Thyroid Associated Ophthalmopathy

Due to the particularity of medical image imaging principle,the complexity of human tissue and the strict requirement of diagnosis result,at present,in clinical medical treatment,it is still dominated by human beings,and computer image technology is supplemented,which also increases the burden of manpower,but also has higher professional and experience requirements,it is this,more automated computer aided diagnosis technology has been the target of the research.there is a lot of work to be done in image processing before the diagnosis of the disease,which is a huge workload for manual,and has a higher precision requirements,automatic image segmentation is also the focus of computer aided diagnosis system.In this paper,the main object is the CT image of the eyeball thyroid.In order to identify whether CT image is abnormal or not,it is necessary for image pre-processing,feature extraction and recognition..First,it can intercept the region of interest as small as possible,and then the data feature is extracted from the segmentation image,and design the classifier to recognize the image.The recognition result is related to image processing effect,feature selection and classifier design.In view of thyroid related ophthalmopathy,this paper considers two aspects of image data preprocessing and image recognition,and makes appropriate pre-processing work according to the characteristics of the image,in order to achieve a better image processing effect.Finally,the image is classified by the existing different algorithms.The main work of this paper is as follows:(1)Image preprocessing is a difficult point in medical image processing,and it is also a key point in computer-aided diagnosis systems.In this paper,a series of pre-processing work on the original CT image data is done,in the process of segmentation of the region of interest,the ratio of effective information in the focus area of the image is improved gradually,and the experimental data samples with different effective information ratio are obtained,and the ratio of different sample experiment recognition rates is compared.(2)According to the characteristics of the data set image,the corresponding image features are selected,the BP neural network and convolution neural network are used to identify the data.The corresponding structure of the corresponding BP neural network and convolution neural network is designed and improved,and the performance of the network is improved,also compare the images with different effective information proportion.