Research On Wisdom Teeth Diagnosis Of Panoramic X-ray Dental Film Based On Machine

Wisdom teeth refer to the third molars in the human oral cavity.Impacted wisdom teeth can bring a lot of trouble,inconvenience and even life-threatening to patients,and are the focus of clinical diagnosis.With the development and application of machine learning and modern evidence-based medicine,more and more diseases have gradually begun to realize or explore the intelligence and automation of the diagnosis process.The key to realize the intelligence and automation of wisdom teeth diagnosis is to use machine learning algorithms to realize the automatic recognition and classification of wisdom teeth.Starting from the classification of wisdom teeth,this paper uses computer graphics and image technology and machine learning technology to study the identification and classification of wisdom teeth.In view of the noise problem of panoramic X-ray dental film,an improved wavelet threshold denoising function is proposed on the basis of the wavelet threshold function.Experiments have verified that the improved threshold function is compared with the traditional wavelet threshold denoising function.The noise effect is more advantageous.Image denoising is achieved by a combination of median filtering and improved wavelet threshold denoising algorithm.According to the needs of making data sets,through designing comparative experiments,the application effects of histogram equalization,area growth method,OTSU method and other methods in wisdom tooth segmentation are analyzed,and the OTSU method is improved and verified,and a fusion is proposed.The segmentation algorithm of histogram equalization,improved OTSU algorithm and digital morphology can effectively extract the second molars and wisdom teeth in panoramic X-ray dental film,and use this method to process panoramic X-ray dental film.According to the needs of traditional machine learning SVM and deep learning convolutional neural network to solve practical problems,the segmented images are processed by feature annotation,etc.,and the corresponding data set is established.Two SVM classification experiments are designed,and five convolutional neural network models are selected for classification experiments.The experimental results show that when the wisdom tooth image sample is small,designing and constructing appropriate image features can improve the classification accuracy of the SVM classification model.Using the CNN network model,you don’t need professionals to perform feature annotation,and you only need to provide enough wisdom tooth images to obtain a model with ideal classification accuracy.Among them,the Vgg Net16 neural network model has the best effect.Finally,based on the shortcomings of the current wisdom tooth diagnosis process,drawing on the theory of narrow evidence-based medicine,combining the proposed improved denoising algorithm and segmentation method,and fusing the Vgg Net16 convolutional neural network model,a reasonable wisdom tooth diagnosis system is designed,and some parts are realized.Features.