Gingivitis Identification via GLCM and Artificial Neural Network

摘要

Gingivitis is a common oral disease. The diagnosis process of gingivi-tis disease is usually based on the experience of the dentist and previous med-ical records. In order to diagnose gingivitis more efficiently and accurately, we proposed a gingivitis recognition program based on Gray-Level Co-Occurrence Matrix (GLCM), Artificial Neural Network (ANN) and Genetic Algorithms (GA). We obtained 180 oral images from Nanjing Stomatological Hospital through two professional medical cameras, 90 of which belong to the gingivitis class, and the rest of which are from the healthy class. We combined GLCM and Artificial Neural Network to identify gingivitis by K-fold Cross-Validation (CV), in our experiment, we utilized the 10-fold Cross-Validation algorithm. We used six evaluation indi-cators to objectively evaluate the classification performance, which are sensitivity, specificity, precision, accuracy, F1, and MCC. Compared with Contrast Limited Adaptive Histogram Equalization (CLAHE) and GLCM plus Extreme Learning Machine (ELM), the identification performance of our algorithm is better than them.