Tooth image caries identification method and system based on convolutional neural network

Abstract

The invention provides a tooth image caries identification method and system based on a convolutional neural network. By performing deep learning on a large number of clinically diagnosed digital photos of teeth, the probability of dental caries in a target photo can be accurately judged. The public can be helped to perform self-supervision of dental caries, and the public can timely go to a hospital to see a doctor for examination when necessary. The method and the system have the characteristics of good real-time performance and capability of discovering early caries.

Description

technical field [0001] The invention relates to a tooth health status judgment system based on deep learning. The judging object is a digital photo. The main problem to be solved is that it is difficult for non-professionals to self-discover early caries. The algorithm can give early warning of early caries. Background technique [0002] Dental caries (dental caries) is a chronic, progressive, destructive disease that occurs in dental hard tissues with bacteria as pathogens and the participation of various factors. Caries is a common and frequently-occurring disease. The World Health Organization lists it as one of the three major diseases that endanger human beings. To detect caries as early as possible, and to prevent and treat them early can prevent The development of dental caries. [0003] Caries, especially children's caries, has a wide range of caries and develops rapidly. The early manifestations of caries are not obvious. When parents find out and seek medical trea...

AI Technical Summary

Problems solved by technology

However, there is no literature report on the use of convolutional neural network and machine learning techniques to identify digital images of proximal caries.

[0007] To sum up, the existing technology has the following technical deficiencies: (1) It needs relatively expensive external imaging hardware, and also has certain requirements on the external environmental conditions, which can be used as a diagnosis aid for medical institutions, but it is not suitable for the daily monitoring of dental caries in the general population. Situation use

(2) At present, the main method for clinical detection of early caries is to require the examinees to go to medical institutions for regular checkups. On the one hand, this method requires sufficient medical resources, and the professional team of stomatologists in my country, especially children's stomatologists, is seriously insufficient; On the one hand, the examinee needs to take time to go to the medical institution, all in all, it needs to consume a lot of social resources, and there will inevitably be waste of both doctors and patients

Summarize the shortcomings of the current method: insufficient resources, waste of resources

(3) There is no precedent for automatically identifying digital photos of proximal caries based on machine learning

[0008] In short, there is currently no method for monitoring early pit and fissure caries (the shape of pit and fissure is relatively complete, manifested as caries at the bottom of the groove wall) and proximal caries (those without destroying the marginal ridge) for popular daily use by the general public. The accessibility of the method is poor, and it needs to consume a lot of doctor-patient resources

Images