Abstract
In this work applied to the field of orthodontics, we study the problem of automatic analysis of 3D dental models. Our goal is to provide solutions to automate the tedious tasks performed by the practitioner when treating a new patient. To do this, we first study the problem of automatic segmentation of 3D dental models. More particularly, we study the means of obtaining a segmentation method that is robust to the diversity of pathological configurations that may be encountered in the field.We then deal with the automatic evaluation of given dental features of the models (such as the orientations of the teeth, or the distance between keypoints), and the automatic arrangement of the teeth on the arches. The approach followed for these two problems is based on the registration of a generic and non-pathological dental model the patient model, and on its deformation. This method allows us to perform automatic measurements in order to evaluate the misalignment of each tooth with respect to a healthy dentition, and to qualify the overall dental situation. We also obtain an estimation of the patient's ideal dentition which can be considered as a treatment goal. This estimated goal can then be adjusted in an interactive system to create a real treatment objective, and produce an animation illustrating the treatment.Finally, the dental situation computed above makes it possible to produce a text reflecting the pathological situation, and which contains the relevant concepts extracted from a knowledge base of the domain. The exploitation of this knowledge makes it possible to decide on a certain number of diagnostic elements necessary to determine the treatment plan.In terms of applications, the work presented can be used to: reduce the time needed to establish a diagnosis (automation of certain measurements and suggestion of elements in natural language), accelerate the creation of a digital treatment objective and produce an animation for the patient, or facilitate the indexing and search of patient models in natural language.