Detection of Dental Caries Using NIR Images

Dental caries is the most prevalent chronic disease, impacting 3.9 billion individuals worldwide. Medical specialists are required for the diagnosis of dental caries. Conventional methods such as bitewing, panoramic, and intraoral radiograph testing are used to detect dental caries. However, frequent and prolonged exposure to X-rays can harm internal organs. In order to overcome this difficulty, a portable, ionization-free near-infrared imaging approach is proposed. The detection of dental caries is automated using several deep-learning algorithms. The main objective of this work is to show how well a cascaded U-Net segmentation technique works in the image acquired from constructed NIR camera (950 nm). The cascaded U-Net model performed better at spotting dental caries with an accuracy of 98.7%.