เทคนิคอัตโนมัติเต็มรูปแบบในการประเมินระยะการพัฒนาของฟันกรามล่างซี่ที่สาม สำหรับการประมาณอายุโดยใช้ตัวตรวจจับคุณลักษณะแบบรวมช่องและเทคนิคการเรียนรู้เชิงลึก

Abstract

Age estimation is an important role in forensic science. It is suggested that radiographic images be utilized to estimate the age of unlive bodies from disasters or crimes. The body may lose vital organs or bones used to calculate age. As a result, estimated ages may be less accurate. Teeth, on the other hand, are organs in the body that rarely lose their information. Demirjian's approach is recommended for estimating age from dental radiography images. Dental age estimation requires a lot of time and knowledge to be done correctly. This study proposes using machine learning to overcome these problems. The model was created by combining aggregate channel features detector and deep learning algorithm. There are three processes. First, use an aggregate channel features detector to locate the lower left third molar. Then, the third molar stages were then evaluate using deep learning. Finally, using the method provided in the previous study, compute the age from the evaluated stages. The experimental results showed that the model can localize the lower left third molar in 99.5% of cases, evaluate the third molar with an accuracy of 83.0% and estimate age with a correlation with coefficient 𝜌 = 0.91 when compared to chronological age. Therefore, this developed model can be used to estimate the age from the dental panoramic films, which can reduce the evaluation time.