Assessment of Reliability of Three Different Computer-Assisted Analysis Programs

Abstract

The aim of this study was to assess the reliability of cephalometric analysis using 3 different digital analysis programs. Methods: For this study, a dry human skull with the entire mandible, maxilla, and teeth was used. Fifteen lateral and 15 posteroanterior cephalometric digital images were taken by rotating the skull from 08 to 6148 at 28 intervals to obtain different images. Two researchers located the landmarks on the digital images independently using 3 computer-assisted analysis programs, Dolphin Image Software (Dolphin Imaging and Management Solutions), Quick Ceph Image (Quick Ceph Systems Inc), and Vistadent OC (GAC Int Inc). Following the first measurements (T1), all landmarks were relocated within a 2-week interval (T2) by each examiner. A paired and the independent Student t tests were used for intraexaminer and interexaminer measurements, and Pearson correlations were obtained. Intraclass correlation coefficients (ICC) were calculated to determine intraexaminer repeatability. Results: For the repeated measurements, mean differences were statistically insignificant, and high correlations for the repeated measurements were found, and the intraexaminer correlations were significant for each examiner (p,.001). When the interexaminer correlations of 3 analyses were compared, interexaminer correlations showed high consistency and the lowest Pearson r value was the same angular measurement (S ant-n-ss) (p,.001). ICC values demonstrated high intraexaminer repeatability. The highest value of ICC was the mandibular body length (go-me) for both examiners (p,.001). Conclusion: The 3 tested analysis programs may be accepted as reliable for clinical use.

A modified Steiner's analysis that does not require radiographic exposure of the cranial base

“…The agreement in landmark identification between the two examiners in this study was higher than that was reported by Uysal et al who had ICC values ranging from 0.484 to 0.727. The inter‐examiner correlations seen in this study were more consistent with those observed by Toy et al who presented correlation values between 0.836 and 0.990 ( P < 0.001 for both studies).…”