Development and evaluation of a standardized method and atlas for contouring primary and permanent dentition

Objectives: Radiation toxicity of the dentition may present significant treatment-related morbidity in the paediatric head and neck cancer population. However, clear dose-effect relationships remain undetermined and must be predicated upon accurate structure delineation and dosimetry at the individual tooth level. Radiation oncologists generally have limited familiarity or experience with relevant dental anatomy. Methods: We therefore developed a detailed CT atlas of permanent and primary dentition. After studying this atlas, five radiation oncology clinicians delineated all teeth for each of eight different cases (selected for breadth of dental maturity and anatomical variability). They were asked to record confidence in their contours on a per-tooth basis as well as the duration of time required per case. Contour accuracy and interclinician variability were assessed by Hausdorff distance and Dice similarity coefficient. All analyses were performed using R v. 3.1.1 and the RadOnc v. 1.0.9 package. Results: Participating clinicians delineated teeth with varying degrees of completeness and accuracy, stratified primarily by the age of the subject. On a per-tooth basis, delineation of permanent dentition was feasible for incisors, canines, premolars and first molars among all subjects, even at the youngest ages. However, delineation of second and third molars was less consistent, commensurate with approximate timing of tooth development. Within each tooth contour, uncertainty was the greatest at the level of the dental roots. Conclusions: Delineation of individual teeth is feasible and serves as a necessary precursor for dental dose assessment and avoidance. Among the paediatric radiation oncology community in particular, this atlas may serve as a useful tool and reference.