A study on the development of a plaque detection program using artificial intelligence

Objectives: To evaluate oral hygiene, disclosing evaluation of dental plaque is a very important diagnostic method. Various indices are applied to the staining and evaluation of these dental plaque, and they are determined by visual inspection by dental experts. In order to develop an image analysis program that can be more helpful in the quantitative evaluation, we tried to devise and evaluate an automatic scoring system by using artificial intelligence in the quantitative evaluation method of dental plaque using tooth staining agent.
Methods: The tooth surface coloring agent used for deep learning training was Monotone (Disclosing solution, THOMASTONE, CO, ltd, KOREA), and 1,000 images were used for learning. Labeling for quantitative evaluation of images was performed by two dental experts, and the dataset was divided into training data, verification data, and test data sets to form an artificial intelligence structure. In order to improve the accuracy of dental plaque, multi-model design and learning were conducted, and step-by-step dental plaque separation and detection results were confirmed.
Results: As a result of calculating and evaluating the degree of disclosing area compared to the teeth of the tooth-colored image using artificial intelligence, A data conversion success rate of over 99% was secured in the parsing process for individual tooth data in tooth photos and the post-parsing DB process, and a normalization process was established for AI analysis in tooth image photos. Also, this program accuracy of 95.5% was shown, and the separation of teeth from the image and the calculation of the tooth-colored range were numerically derived.
Conclusions: It was able to derive that the introduction of artificial intelligence in dental plaque color image analysis to evaluate oral hygiene is highly applicable.