Comparing Visualization Methods of Cone Beam Data to Assess A Mandibular Third Molar Development Classification System

Abstract

Purpose: To visualize third molar development staging, created by Dr. Christine Hammer in 2015, from three different image types extracted from CBCT data in order to determine if there is a difference between visualization methods in reliably categorizing mandibular third molars. Specific Aims: To evaluate the level of agreement and reliability amongst three raters utilizing Hammer’s 8 stage classification system of third molar development formulated in 2015 with additional visualization methods extracted from CBCT data; To compare three visualization methods—3D semi-transparent volume, corrected sagittal slice, 2D extracted panoramic radiograph— and determine if they differ in their reliability for classifying third molars into eight developmental stages. Hypothesis: All kappa values will be equal to zero, and there will be a low level of inter-rater agreement in mandibular third molar staging for all image types; there will be no difference between three different visualization methods of Hammer’s 8 stage classification system for third molar development. Methods: Sixty total subjects who are chronologically between the ages of 10-18 were included and divided into three equal subgroups: 10-12 years old, 13-15 years old, and 16-18 years old with 20 subjects each. For each subject, three different radiographic images of one left mandibular third molar were analyzed. Three raters, an orthodontic resident, oral and maxillofacial radiologist, and orthodontist, evaluated each image. A calibration meeting was held to appropriately stage example mandibular third molar images with Dr. Christine Hammer’s stage I-VIII classification system. For the study, 180 total images (60 subjects with 3 images each) were rated by each rater individually, and 90 of those images were randomly selected and re-rated 5 days later.Intra- and inter- rater agreement for Hammer’s staging of mandibular molars was assessed using Kappa statistics. Kappa values were reported for inter-rater within each of the three imaging modalities (3D semi-transparent volume, corrected sagittal slice, and extracted 2D panoramic). Pairwise comparison between imaging modalities were conducted using Cohen’s Kappa to assess intra-rater reliability for each rater. Results: The null hypothesis was rejected. There was a high level of agreement amongst raters, and all raters were reliable in identifying Hammer’s 8 stages. Rater 1 showed the most intra-rater reliability overall. There was also a difference amongst image types: Sagittal slices extracted from the CBCT showed the highest percentage of agreement amongst all ages and all image types, thus having the greatest inter-rater reliability. 3D images had the lowest inter-rater reliability, with panoramic images being second behind slices. Slices also had the greatest intra-rater reliability for all ages and categories, except for 10-12 year olds, in which panoramics had the highest. Conclusions: Dental professionals can reliably utilize Hammer’s third molar classification system to stage development. Sagittal slices extracted from a CBCT have the most overall inter-rater and intra-rater agreement. These appear to be the most reliable modality when wanting to view tooth development and tooth staging. Despite this, traditional 2D panoramic images can still be reliably used by clinicians to analyze third molar development. Hammer’s 8 stage classification is a reliable tool for determining mandibular third molar classification, with the hardest stages for raters to determine was the difference between stages 3 and 4 and between stages 5 and 6. Overall, acquiring a 3D image and 3D data can be beneficial for identifying mandibular third molar classification, if the clinician is willing to purchase a CBCT system and dedicate time to analyze the data via slices.