Date of Award
Spring 1-1-2013
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Dentistry
First Advisor
Bosio, Jose A.
Second Advisor
Bradley, T. Gerard
Third Advisor
Liu, Dawei
Abstract
Introduction:With attention being given to the deleterious effects of radiation exposure from dental radiographs and inaccuracies in cephalometric soft tissue measurements, an alternative method of facial analysis with sufficiently reliable soft tissue landmarks should be developed. The goals of this study were threefold: (1) to define a new, low-cost method for capturing standardized frontal and sagittal facial images, (2) to determine on which photographic view that landmarks can be more reliably located, and (3) to determine which landmarks are appropriate for quantitative facial analysis.
Materials and Methods:Simultaneous frontal and right sagittal facial images of 10 male and 10 female dental student subjects were captured using high-definition webcams as part of a low-cost set-up. Seventeen identical predefined facial soft tissue landmarks were located by 5 examiners on both types of images and were recorded as coordinate values. These coordinate values were used to calculate the best estimate of the true value for each landmark, mean deviation from this best estimate, and reliability in the X- and Y-axes using the Shrout-Fleiss intraclass correlation coefficient with corresponding 95% confidence intervals. Two examiners repeated the landmark location to evaluate intra-examiner reliability.
Results:With a 95% confidence interval range of >0.950, nose and mouth landmarks were among the most reliable landmarks on frontal and sagittal facial images. Converselyright soft tissue gonionwas one of the least reliable landmarks located in this study. In general, landmarks located by a single examiner showed greater reliability than when there were multiple examiners.
Conclusions:This low-cost method yielded frontal and sagittal images sufficient for landmark identification. The magnitude of error varies between landmarks, is largest for poorly demarcated landmarks, and most had a non-circular envelope of error. Certain landmarks were more reliable on sagittal images and others were more reliable on frontal images. All landmarks had greater reliability and less mean deviation when located by a single examiner.