Comparison of W. Arnett's cephalometric analysis measurements performed using artificial intelligence and manual measurements.

Clinical Snapshot

Key Findings

• P Value: Not specified for primary comparison

• Effect Size: ICC values ranging 0.913-0.995 for software vs manual comparison

• Primary Outcome: Agreement between WebCeph Premium™ and manual cephalometric measurements

• Nnt Or Sensitivity: High agreement with ICC >0.9 for most measurements, indicating excellent reliability

• Confidence Interval: Not reported

Clinical Application

Requires access to WebCeph Premium™ software and appropriate training. Cost-effectiveness not evaluated Relevant to Australian orthodontic practice where cephalometric analysis is standard. Software would need TGA approval if classified as medical device. Cost considerations important for private practice adoption Orthodontic patients requiring cephalometric analysis for treatment planning

Abstract

OBJECTIVES: This study aimed to compare the linear and angular Arnett cephalometric measurements obtained from WebCeph Premium™ with those from manual cephalometric analysis and to assess the reliability of the automated measurements. METHODS: Thirty-two pre-treatment lateral cephalograms of patients were randomly selected. Images were calibrated, 29 landmarks were manually traced, and 40 parameters were recorded by two orthodontists (6 angular, 34 linear). WebCeph Premium™ (version 2.0.0, AssembleCircle Corp., Gyeonggi-do, Republic of Korea) automatically identified landmarks and performed Arnett's cephalometric analysis. After 15 days, orthodontists reassessed the radiographs. Reliability and repeatability were evaluated using the intraclass correlation coefficient (ICC), which exceeded 0.95, indicating excellent agreement. Normality was assessed using the Shapiro-Wilk test, and group comparisons were performed with paired t-tests. RESULTS: For Researcher 1, initial ICC values ranged from 0.003 to 0.984 and final values from 0.617 to 0.996. For Researcher 2, initial ICC values ranged from 0.022 to 0.999 and final values from 0.601 to 0.999. Inter-observer ICC values ranged from 0.03 to 0.984, with most measurements showing high agreement. Manual tracing and WebCeph Premium™ showed high agreement, with ICC values ranging 0.913 and 0.995. CONCLUSIONS: WebCeph Premium™ demonstrates high concordance with manual Arnett analyses; however, further refinement of certain parameters is necessary to enhance AI precision. CLINICAL RELEVANCE: The utilization of WebCeph Premium™ within clinical settings has the capacity to enhance the efficiency of cephalometric analysis for orthognathic cases by minimizing manual workload and inter-examiner variability. This can result in more consistent and efficient treatment planning.

References

1. 1.Kula, B., & Polat, E. N. A. (2026). Comparison of W. Arnett's cephalometric analysis measurements performed using artificial intelligence and manual measurements. Clinical Oral Investigations. https://doi.org/10.1007/s00784-021-03990-w