Thermographic Toothache Screening by Artificial Intelligence.

Inflammatory toothache is a painful condition frequently reported by patients in the daily dental clinic, and may reach a disabling situation, directly impacting their quality of life and systemic health. Panoramic and periapical radiographies can demonstrate periapical pathologies and lesions of the dental organ, but cannot record the local sympathetic vasomotor response. Infrared thermography is a functional imaging method with no emission of ionizing radiation, it is quick to acquire, and reflects the microcirculatory dynamics of the skin surface, enlarging the vascular, nervous, and skeletal muscle, as well as inflammatory processes. The study tested the application of four models of the artificial neural network (MobileNetV2, InceptionV3, ResNet101V2, and ResNet50V2) in the detection of inflammatory toothache. The sample consisted of 76 volunteers, over 18 years old, being selected 51 volunteers with inflammatory toothache and 25 volunteers compatible with normality. They were submitted to thermographic, panoramic radiographic and clinical examinations. Among the four models tested, MobileNetV2 performed better in both front and lateral views. However, considering the ROC curve, the best performance of the automated detection of toothache cases was by the frontal image. The authors discussed how this technique can help the diagnostic accuracy in dentistry and the future perspective of artificial intelligence in the area of endodontics. Infrared thermography can provide health professionals with objective and accurate information about the vasomotor and facial neurovegetative responses in inflammatory toothache. It may assist in the screening of volunteers with inflammatory toothache using artificial neural networks.