سه شنبه 2 تیر 1405
دوره 14، شماره 2 - ( 2-1405 )
جلد 14 شماره 2 صفحات 154-145 |
برگشت به فهرست نسخه ها
Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
Emadi Majd F, Langari S F, Langari S H. Artificial Intelligence in Pediatric Lateral Cephalometric Detection and Analysis: A Narrative Review. J. Pediatr. Rev 2026; 14 (2) :145-154
URL: http://jpr.mazums.ac.ir/article-1-819-fa.html
URL: http://jpr.mazums.ac.ir/article-1-819-fa.html
Artificial Intelligence in Pediatric Lateral Cephalometric Detection and Analysis: A Narrative Review. Journal of Pediatrics Review. 1405; 14 (2) :145-154
چکیده: (4 مشاهده)
Background: Malocclusion is a common condition in children, often requiring timely orthodontic intervention during critical stages of craniofacial development. Lateral cephalometric radiography remains a fundamental diagnostic modality; however, manual landmark identification suffers from subjectivity, time consumption, and inter-examiner variability—challenges that are amplified in growing pediatric patients.
Objectives: To review recent applications of artificial intelligence in pediatric lateral cephalometric analysis, focusing on automated landmark detection, diagnostic accuracy, clinical usefulness, educational value, and radiation-reducing innovations.
Methods: This narrative review synthesizes peer-reviewed literature published between January 2020 and October 2025. Studies evaluating artificial intelligence (AI) systems for cephalometric analysis in pediatric or adolescent populations were included, with emphasis on methodological approaches, accuracy metrics, clinical feasibility, and innovations such as low-dose imaging or photograph-based landmark prediction.
Results: Deep learning models—particularly convolutional and region-based neural networks—consistently achieved landmark localization errors below 2 mm, meeting clinically acceptable thresholds. AI integration significantly reduced analysis time (from 10–30 minutes to seconds) and enhanced reproducibility, especially when used in hybrid workflows combining algorithmic output with clinician verification. Educational applications showed improved trainee performance, while emerging non-ionizing approaches (e.g. AI-driven prediction from facial photographs) offered promising radiation-reduction strategies. Nevertheless, limitations persist, including platform-specific performance disparities, insufficient diversity in training datasets, and limited model interpretability.
Conclusions: AI is not a replacement for clinical expertise but a powerful adjunct in pediatric cephalometric analysis. It enhances diagnostic efficiency, supports education, and enables safer imaging alternatives. Future implementation must prioritize diverse pediatric data, transparent and explainable architectures, and real-world validation to ensure equitable, reliable, and ethically sound clinical adoption.
Objectives: To review recent applications of artificial intelligence in pediatric lateral cephalometric analysis, focusing on automated landmark detection, diagnostic accuracy, clinical usefulness, educational value, and radiation-reducing innovations.
Methods: This narrative review synthesizes peer-reviewed literature published between January 2020 and October 2025. Studies evaluating artificial intelligence (AI) systems for cephalometric analysis in pediatric or adolescent populations were included, with emphasis on methodological approaches, accuracy metrics, clinical feasibility, and innovations such as low-dose imaging or photograph-based landmark prediction.
Results: Deep learning models—particularly convolutional and region-based neural networks—consistently achieved landmark localization errors below 2 mm, meeting clinically acceptable thresholds. AI integration significantly reduced analysis time (from 10–30 minutes to seconds) and enhanced reproducibility, especially when used in hybrid workflows combining algorithmic output with clinician verification. Educational applications showed improved trainee performance, while emerging non-ionizing approaches (e.g. AI-driven prediction from facial photographs) offered promising radiation-reduction strategies. Nevertheless, limitations persist, including platform-specific performance disparities, insufficient diversity in training datasets, and limited model interpretability.
Conclusions: AI is not a replacement for clinical expertise but a powerful adjunct in pediatric cephalometric analysis. It enhances diagnostic efficiency, supports education, and enables safer imaging alternatives. Future implementation must prioritize diverse pediatric data, transparent and explainable architectures, and real-world validation to ensure equitable, reliable, and ethically sound clinical adoption.
نوع مطالعه: Narrative Review |
دریافت: 1404/8/10 | پذیرش: 1404/8/19 | انتشار: 1405/1/23
دریافت: 1404/8/10 | پذیرش: 1404/8/19 | انتشار: 1405/1/23
| بازنشر اطلاعات | |
 |
این مقاله تحت شرایط Creative Commons Attribution-NonCommercial 4.0 International License قابل بازنشر است. |
