The aim of this project is to evaluate the feasibility of artificial intelligence (AI) assisted B-mode ultrasound based digital monitoring to assess buccal peri-implant tissues in an ex-vivo porcine model, using parameters that enable a future clinical trial. The clinical relevance of this study is to create future conditions for any dentist to accurately measure peri-implant tissues with ultrasound, even without experience with this technology. The difficulty would be the evaluation of the ultrasound image. Therefore, the aim of this project is to create a tool based on AI to facilitate the measurements. The AI-based method is expected to help robustly segment the anatomical structures after US scanning, co-registration with a-priori information and evaluation in comparison with earlier clinical data, save time and making the ultrasound technology more user-friendly and feasible in clinical practice. The hypothesis is that B-mode ultrasound measurements present agreement with cone-beam computed tomography and direct measurements in a way it can be utilized as a supplementary tool to monitor peri-implant tissue dimensions in daily clinical routine. For the clinical assessment of buccal bone and mucosal thickness, only 2D distances have to be measured at defined positions. 3D ultrasound-based reconstructions add just little value but requires a great deal of equipment. In contrast, the use of B-mode ultrasound is widespread, easier to apply, and in recent years the signal-to-noise ratio has greatly improved. B-mode ultrasound probes offer the advantage of being much smaller than 3D probes and thus greatly simplify intraoral buccal measurement. However, unlike 3D ultrasound, B-mode ultrasound provides only a single (2D) image. Whereas this can be sufficient for 2D monitoring of peri-implant mucosa thickness, solutions for 3D-registration with the crown-implant construct and a reproducible and intuitive alignment of the ultrasound-probe need to be developed and evaluated. Thus, two further hypotheses have to be evaluated: (1) An accurate and robust registration process can be achieved using B-Mode US; and (2) A fully automatic robust processing and measurement can be achieved using AI.
DFG Programme
Research Grants
International Connection
Austria
