A Comparative Study of CNN and DBN Algorithms in Oral Disease Imaging Diagnosis

Abstract

Oral imaging involves imaging techniques and diagnostic methods of the internal structure of the oral cavity. Intelligent diagnostic techniques based on machine learning and medical gold standards have become a research hotspot in recent years. In order to obtain the accuracy of deep learning for oral image segmentation, this paper mainly compares the performance of deep belief network (DBN) and convolutional neural network (CNN) in image segmentation and recognition. Firstly, the CBCT image data set of the teeth of the relevant cases was established, and the image was preprocessed and marked, and the training set, test set and verification set were divided. Secondly, the DBN and CNN deep learning neural network models are built and debugged for training and testing respectively. Finally, the performance of the two models is compared and their advantages and disadvantages are analyzed from the data of graphic distortion, feature extraction accuracy, tooth discrimination and accuracy. The experimental results show that both of them have good performance in image segmentation, but the CNN model is superior to DBN in oral image classification and recognition. This study explores the application of deep learning in CBCT image processing of teeth. The research results are of positive significance for assisting doctors to judge the condition more accurately, improving the treatment effect and quality, and improving the digitization and intelligence of oral disease imaging diagnosis.