การจำแนกภาวะฟันตกกระโดยใช้ขั้นตอนวิธีการจัดกลุ่มแบบฟัซซีด้วยควอนตัมอินสไปร์

Abstract

Dental fluorosis occurs when a person is exposed to a lot of fluoride during the teeth developing period when their teeth are developing. This phenomenon can be found all over the world especially in water resource area with highly fluoride. This causes the teeth to lose an enamel layer, affecting the beauty of the teeth to the point of causing the teeth to become porous and less strong than usual. Therefore, the study of dental fluorosis using a nature-inspired fuzzy clustering algorithm is intended to help dentists to save their time in diagnosis dental fluorosis. There are 4 levels of dental fluorosis considered in this thesis, i.e., normal teeth. Fractured Tooth Level 1 Fractured Tooth Level 2 Fractured Tooth Level 3 according to the Dean’s index [5]. The images are divided into 4 groups of pixels: gums and lips, normal white teeth, dental fluorosis white and dental fluorosis brown. After that, each pixel types are grouped using the Enhance Quantum-Inspired Evolutionary Fuzzy C-Means (EQIE-FCM) [6-8] to group and find prototypes. Then, K-Nearest Neighbors is used to classify each pixel of images. Finally, the Dean’s index is used to determine level of dental fluorosis. According to the results of the study, the average Intersection over Union were 0.475 for automatic gum-teeth separation and 0.64 for manual gum-teeth separation. However, the best result from automatic classification is approximately 33.63% whereas that of the manual system is around 42.48%. The reason for the wrong classification might be because the reflection in the image, shadow in the image, and dirt stain in the image.