Medical Image Noise Reduction And Computer-aided Detection

With the advances in science and the rapid development of computer technology,the life of people is slowly transforming to the intelligence.Especially in the field of medicine,more and more advanced instruments and equipment are used in various medical and clinical diagnosis.Due to the influence of medical imaging equipment and the complexity of human tissue,the original medical images of the human tissue which were collected through medical imaging equipment is low in Signal to Noise Ratio.This images must though a certain image processing before it can be as a medical diagnosis reference images.Therefore,medical image noise reduction technology and computer aided diagnosis have received more and more attention.In addition,the performance of noise reduction technology will influence the results of computer aided diagnosis which is importance for judgment of the lesions.In the process of obtaining the medical images will produce a certain amount of radiation,radiation will cause a certain degree of harm to doctors and patients,which is likely to cause cancer and other diseases.Therefore,low dose computed tomography(CT)scanning technology obtains more attention of researchers in recent years.However,with the reduction of radiation dose,the medical image will introduce a lot of noise.Hence,medical noise reduction technology requires more and more attention of scholars.Meanwhile,the doctor often need to rely on the clinical experience and academic background knowledge when doctor determines the lesions in the process of the follow-up.Therefore,it is difficult to gain popularity to determine lesions.Base on this basics,in this paper,the medical image noise reduction processing technology and computer aided diagnosis are made a more in-depth research.In addition,this paper highlights the medical image noise reduction method and the computer aided diagnosis of mammary gland calcify point method.The specific content is as follows:(1)An adaptive three-dimensional medical image noise reduction method which is based on a no reference medical image quality assessment is proposed in this paper.Because the original medical images usually contain noise,the different threshold value of three-dimensional block matching is employed to reduce the the noise.Next,the image quality of the images after noise reduction should be evaluated to obtain the best quality image.Finally,the image with best quality and the corresponding threshold will be output.The experimental results show that this algorithm can effectively suppress noise while keep the details of images better.(2)Rotation invariant local binary patterns is used to extract the texture features of mammary gland calcify point in this paper.In addition,machine learning is utilized to realize the early mammary gland calcify point detection.Calcified points is one form of early diagnosis of breast cancer,because the texture of breast tissue varies according to whether mammary gland contains calcified points.The rotation invariant local binary patterns is often used to extract the texture feature information of images,so it can be used to extract the texture feature of breast tissue.Next,a machine learning technology can be used to divide the extracted features into two categories to determine whether breast tissue containing calcified points.