A scoping review of twenty studies finds artificial intelligence demonstrates high diagnostic accuracy in forensic dental imaging but requires human oversight due to methodological limitations.
Overview of AI in Forensic Odontology A recently published scoping review in BMC Medical Imaging reveals that deep learning models exhibit high diagnostic accuracy in forensic dental imaging. However, the review emphasizes that these AI tools currently necessitate human oversight due to inherent methodological limitations. The growing interest in artificial intelligence in this field is driven by its potential to introduce greater objectivity, enhance efficiency, and improve the repeatability of forensic analyses. Despite these benefits, the authors conclude that automated algorithms should primarily function as decision-support tools, with human specialists retaining ultimate control. Review Scope, Objectives, and Methodology Conducted by Gurbet Alev Oztas Sahiner and Bengisu Karatas, this comprehensive scoping review aimed to delineate the applications of artificial intelligence within forensic odontology. The investigation specifically evaluated AI’s utility across key forensic tasks, including dental age estimation, biological sex estimation, human identification, and tooth classification. Additionally, the researchers explored recent multimodal artificial intelligence applications in forensic dentistry, providing insights into their diagnostic performance, identifying existing methodological constraints, and outlining crucial areas for future research. The review meticulously followed PRISMA-ScR guidelines, involving a systematic search across three prominent electronic literature databases: PubMed/MEDLINE, Scopus, and Google Scholar, for relevant studies published between January 2022 and December 2025. The search strategy combined targeted keywords with explicit inclusion and exclusion criteria to ensure the identification of pertinent original research. Study Inclusion, Data Analysis, and Key Applications The review incorporated original studies encompassing various artificial intelligence methodologies, including machine learning, deep learning, hybrid models, and multimodal large language models, all applied to forensic dentistry. Two independent reviewers diligently performed duplicate study selection, data extraction, and a formal methodological quality assessment utilizing the QUADAS-AI tool. Given the significant methodological heterogeneity observed across the selected literature, the authors opted for a narrative synthesis rather than a quantitative statistical meta-analysis. From the initial search, a total of twenty articles met all predefined inclusion criteria. Among the various applications analyzed, dental age estimation emerged as the most frequently studied forensic problem, followed by human identification, biological sex estimation, and tooth classification, in decreasing order of research focus. This highlights the areas where AI research in forensic odontology has been most concentrated. AI Performance and Methodological Gaps The study found that deep learning architectures, particularly convolutional neural networks, consistently demonstrated high diagnostic accuracy across multiple dental imaging modalities. These networks proved effective in processing diverse two-dimensional and three-dimensional radiographic records commonly used in dental forensic investigations, including panoramic radiographs, cone-beam computed tomography images, intraoral scans, and cephalometric radiographs. However, despite these positive diagnostic results, the evidence base revealed several notable methodological weaknesses. The researchers underscored limitations such as retrospective study designs, reliance on patient-specific sample sets, a widespread lack of external validation, and high methodological heterogeneity among the included studies. Consequently, while artificial intelligence offers substantial potential to enhance standardization, efficiency, and objectivity in forensic analysis, the authors firmly conclude that it must currently function strictly as a decision-support aid, with continuous human oversight being essential.
Overview of AI in Forensic Odontology
A recently published scoping review in BMC Medical Imaging reveals that deep learning models exhibit high diagnostic accuracy in forensic dental imaging. However, the review emphasizes that these AI tools currently necessitate human oversight due to inherent methodological limitations. The growing interest in artificial intelligence in this field is driven by its potential to introduce greater objectivity, enhance efficiency, and improve the repeatability of forensic analyses. Despite these benefits, the authors conclude that automated algorithms should primarily function as decision-support tools, with human specialists retaining ultimate control.
Review Scope, Objectives, and Methodology
Conducted by Gurbet Alev Oztas Sahiner and Bengisu Karatas, this comprehensive scoping review aimed to delineate the applications of artificial intelligence within forensic odontology. The investigation specifically evaluated AI’s utility across key forensic tasks, including dental age estimation, biological sex estimation, human identification, and tooth classification. Additionally, the researchers explored recent multimodal artificial intelligence applications in forensic dentistry, providing insights into their diagnostic performance, identifying existing methodological constraints, and outlining crucial areas for future research. The review meticulously followed PRISMA-ScR guidelines, involving a systematic search across three prominent electronic literature databases: PubMed/MEDLINE, Scopus, and Google Scholar, for relevant studies published between January 2022 and December 2025. The search strategy combined targeted keywords with explicit inclusion and exclusion criteria to ensure the identification of pertinent original research.
Study Inclusion, Data Analysis, and Key Applications
The review incorporated original studies encompassing various artificial intelligence methodologies, including machine learning, deep learning, hybrid models, and multimodal large language models, all applied to forensic dentistry. Two independent reviewers diligently performed duplicate study selection, data extraction, and a formal methodological quality assessment utilizing the QUADAS-AI tool. Given the significant methodological heterogeneity observed across the selected literature, the authors opted for a narrative synthesis rather than a quantitative statistical meta-analysis. From the initial search, a total of twenty articles met all predefined inclusion criteria. Among the various applications analyzed, dental age estimation emerged as the most frequently studied forensic problem, followed by human identification, biological sex estimation, and tooth classification, in decreasing order of research focus. This highlights the areas where AI research in forensic odontology has been most concentrated.
AI Performance and Methodological Gaps
The study found that deep learning architectures, particularly convolutional neural networks, consistently demonstrated high diagnostic accuracy across multiple dental imaging modalities. These networks proved effective in processing diverse two-dimensional and three-dimensional radiographic records commonly used in dental forensic investigations, including panoramic radiographs, cone-beam computed tomography images, intraoral scans, and cephalometric radiographs. However, despite these positive diagnostic results, the evidence base revealed several notable methodological weaknesses. The researchers underscored limitations such as retrospective study designs, reliance on patient-specific sample sets, a widespread lack of external validation, and high methodological heterogeneity among the included studies. Consequently, while artificial intelligence offers substantial potential to enhance standardization, efficiency, and objectivity in forensic analysis, the authors firmly conclude that it must currently function strictly as a decision-support aid, with continuous human oversight being essential.