Evaluation of AI model for cephalometriclandmark classi�cation (TG Dental)

Schwendicke Falk,Chaurasia Akhilanand,Tanne Johannes,†. Krois,Behnaz Mohammad,Motamedian Saeed,Mohammad-Rahimi Hossein,Haiat Anahita,Vinayahalingam Shankeeth

Abstract:. Abstract Purpose: The accuracy of cephalometric landmark identification for malocclusion classification is essential for the diagnosis and treatment planning. The identification of these landmarks is often complex and time consuming for the orthodontists. An AI model for the classification was recently developed. Due to the strict regulation on software systems but few content on AI requirements in this publication the model was investigated under consideration of current regulatory considerations. Methods: The AI platform of the ITU/WHO is used to allow the assessment of the models of the application. An Audit was performed assessing the development process with regard to medical device regulations, data protection regulation and ethical consideration. Upon that the major task during the development were evaluated such as qualification, annotation procedure and data set attributes. Results: The AI models were investigated under consideration of technical, clinical, regulatory and ethical consideration. The risk to the health of the patient and user can be considered as low according to the IMDRF definition. Conclusion: The application is useful to aid the decison and treatment planning for malocclusion classification on lateral cephalograms without cephalometric landmarks. It is comparable with common standards in orthodontic diagnosis.

Medicine,Computer Science