Employing Artificial Intelligence for Non-Invasive Diagnosis of Craniofacial Deformities: A Systematic Review of Non-Ionizing Radiation Methods

Introduction: Craniofacial deformities, like craniosynostosis and plagiocephaly, require early diagnosis, but traditional methods like CT scans use harmful ionizing radiation. Artificial Intelligence (AI) offers a non-invasive, radiation-free alternative, using 2D and 3D imaging for accurate diagnosis, improving safety and accessibility, especially in remote areas. Objective: This review evaluates the use of AI in diagnosing craniofacial deformities without ionizing radiation, focusing on its strengths, limitations, and challenges in clinical implementation. Methods: The literature was searched across five databases—PubMed, Google Scholar, ScienceDirect, Wiley, and Cochrane—over the last decade, following PRISMA guidelines. Inclusion criteria were AI-based, non-invasive imaging techniques for craniofacial deformities. The analysis reviewed diagnostic methods, accuracy, types of deformities, data sources, risk of bias, and external validity. Results: Of 166 articles, 12 met the inclusion criteria and involved over 5,000 individuals. AI diagnostic accuracy ranged from 70.2% to 99.5%. Machine learning methods using 2D and 3D images effectively diagnosed conditions like craniosynostosis and plagiocephaly without ionizing radiation exposure. Conclusion: AI provides a safe, effective method for diagnosing craniofacial deformities without ionizing radiation. Further research and Development are necessary for wider clinical adoption.