A Spiking Neural Network for Tooth Chromaticity Detection

Abstract

In recent years, traditional tooth Chromaticity detection technology has been unable to meet the needs of intelligence and efficiency. In this paper, we propose a spiking neural network model for tooth chromaticity detection. First, a training method of Tempotron supervised learning rules combined with linear decay weight momentum is proposed. Aiming at different training stages, the process of the network weight update is improved. Second, a data set specifically for tooth chromaticity detection is established. According to the common dental chromatic lesions in clinical dentistry, such as fluorosis and tetracycline teeth, the chromaticity of teeth is divided into four categories. Experimental results using self-built datasets show that the task accuracy of the network is as high as 96.67%, and the convergence speed of the network has also been significantly improved.