Artificial Intelligence for anatomical segmentation and use cases in CBCTs

Abstract

Introduction: Cone beam CTs are generally taken for treatment planning in major dental treatments. A large number of use cases revolve around the detection, segmentation and measurements between different anatomical structures. The output of separate stl segmentations for each anatomical structure from an artificial intelligence model allows for vastly improved visualisation and measurements which aids in dental treatment planning. Case Description: A world's first deep learning model (artificial intelligence model) has been developed to detect and segment a cone beam CT into up to 135 different segmentations in under two and a half minutes. This type of granular segmentation will aid in all areas of dentistry, including examination, diagnosis, treatment planning and patient education.In this case study, these granular segmentations have been used to highlight three uses cases for the anatomical output: Root canal treatments, implant planning and lower third molar extraction. Using the AI models, automatic visualisation of the treatment areas as well as automatic measurements were provided for clinical use. Discussion: The segmentations were made as granular as possible, to be able to provide the clinician and patient as much information as possible to aid in treatment planning. Examples of this are outlined in the three use cases highlighted in this case report. The detection of different layers of bone and tooth allow for automatic measurements while allowing more precise and efficient treatment planning. Conclusion/Clinical Significance: This new technology will aid clinicians make better and more efficient clinical decisions in wide range of use cases.