Consistency of Velar Shape Identification using Lateral Cephalogram

Objectives: In order to efficiently evaluate velar shape for various fields – obstructive sleep-apnea, cleft palate management or speech function and upper airway structures investigation, the reproducibility of soft palate shape identification on lateral cephalometric radiographs in various groups of reader is assessed.
Methods: One hundred and sixty eight pairs of lateral cephalographs acquired by 2 types of digital cephalomgraphic machines were included for the study. The study inclusion criteria included diagnostic quality of soft palate and at least one available replicated radiograph within the same visit. Demographic data including age and sex, in addition to, postural and structural changes in retake images were recorded. Three groups of reader: 2 experienced oral and maxillofacial radiologists, 2 general dentists and 3 dental students classified soft palate using the velar shapes from You and colleagues’ classification. Repeated examination of 20 cases was done with at least 1 week interval. Kappa analysis was calculated for intra- and inter-observer agreement.
Results: The age of subjects in this study ranges between 7 and 67 years old (22.14+10.1), consisting of 93 female and 77 male. Average percent velar classification in the image-pair are rat-tail(33.0-33.8%), leaf-shape(41.2%), butt-like(8.2-9.3%), straight-like(8.6-8.7%), S-shape (1.5-1.9%) and hook shape(6.1-6.7%). Overall, there was fair agreement between image pair (score 0.39, p<0.00): moderate in radiologist group (score 0.45, p<0.00), fair in GP and DDS groups (score 0.34 and 0.36, p<0.00) or fair in one shot machine group (score 0.35, p<0.00) and moderate in scanning machine group (score 0.51, p<0.00).
Conclusions: Velar shape identification using cephalogram have only fair to moderate reliability.