Evaluating the accuracy of automated cephalometric analysis based on artificial intelligence

Abstract

Background The purpose of this study was to evaluate the accuracy of automatic cephalometric landmark localization and measurements using cephalometric analysis via artificial intelligence (AI) compared with computer-assisted manual analysis. Methods Reconstructed lateral cephalograms (RLCs) from cone-beam computed tomography (CBCT) in 85 patients were selected. Computer-assisted manual analysis (Dolphin Imaging 11.9) and AI automatic analysis (Planmeca Romexis 6.2) were used to locate 19 landmarks and obtain 23 measurements. Mean radial error (MRE) and successful detection rate (SDR) values were calculated to assess the accuracy of automatic landmark digitization. Paired t tests and Bland‒Altman plots were used to compare the differences and consistencies in cephalometric measurements between manual and automatic analysis programs. Results The MRE for 19 cephalometric landmarks was 2.07 ± 1.35 mm with the automatic program. The average SDR within 1 mm, 2 mm, 2.5 mm, 3 and 4 mm were 18.82%, 58.58%, 71.70%, 82.04% and 91.39%, respectively. Soft tissue landmarks (1.54 ± 0.85 mm) had the most consistency, while dental landmarks (2.37 ± 1.55 mm) had the most variation. In total, 15 out of 23 measurements were within the clinically acceptable level of accuracy, 2 mm or 2°. The rates of consistency within the 95% limits of agreement were all above 90% for all measurement parameters. Conclusion Automatic analysis software collects cephalometric measurements almost effectively enough to be acceptable in clinical work. Nevertheless, automatic cephalometry is not capable of completely replacing manual tracing. Additional manual supervision and adjustment for automatic programs can increase accuracy and efficiency.

Correlation Analysis of Nasal Septum Deviation and Results of AI-Driven Automated 3D Cephalometric Analysis

“…Similar results were achieved by other research teams, demonstrating a superior AI success classification rate compared to humans in some cephalometric analysis measures [30,31,45]. An interesting study by Bao et al (2023) evaluated the accuracy of AI in the automated cephalometric analysis of reconstructed lateral cephalograms from CBCTs for 85 patients. The mean radial error for 19 chosen landmarks was 2.07 ± 1.35 mm and an error of less than 3 mm in 71.7% with the automatic program.…”

“…The mean radial error for 19 chosen landmarks was 2.07 ± 1.35 mm and an error of less than 3 mm in 71.7% with the automatic program. The authors concluded that automatic analysis is almost effective enough to be acceptable in clinical work, but is not currently capable of completely replacing manual tracing [46]. Some minor inaccuracies have also been found in previously published papers regarding the reliability of CephX cephalometric analysis.…”

Influence of growth structures and fixed appliances on automated cephalometric landmark recognition with a customized convolutional neural network

“…Finally, since the reference points are used in a further step of the cephalometric analysis to perform angular measurements [ 52 ], a potential limitation of the proposed framework is that such measurements and index data were not obtained. Nevertheless, it should be noted that the cephalometric angles depend to a large extent on the correct positioning of the reference points.…”

Application of Artificial Intelligence in Orthodontics: Current State and Future Perspectives

“…In general, acceptable linear and angular measurement are less than 2 mm and 2 • , respectively [23,36,38,43,44,47,54,[66][67][68][69][70]. Following this criterion, although some commercially available software can achieve high overall accuracy in automated landmarking on lateral cephalograms, manual supervision is still recommended [47,48,53,54,56].…”