Deep learning for forensic age estimation using orthopantomograms in children, adolescents, and young adults

Teeth are known to be the most accurate age indicators of human body and are frequently applied in forensic age estimation. We aimed to validate data mining-based dental age estimation, by comparing the accuracy of the estimation and classification performance of 18-year thresholds with conventional methods and with data mining-based age estimation. A total of 2657 panoramic radiographs were collected from Koreans and Japanese populations aged 15 to 23 years. They were subdivided into a training and internal test set of 900 radiographs each from Koreans, and an external test set of 857 radiographs from Japanese. We compared the accuracy and classification performance of the test sets from conventional methods with those from the data mining models. The accuracy of the conventional method with the internal test set was slightly higher than that of the data mining models, with a slight difference (mean absolute error < 0.21 years, root mean square error < 0.24 years). The classification performance of the 18-year threshold was also similar between the conventional method and the data mining models. Thus, conventional methods can be replaced by data mining models in forensic age estimation using second and third molar maturity of Korean juveniles and young adults.

Forensic age estimation from aspartic acid racemization (AAR): Scrutiny of multiple source samples

Introduction: Forensic dentistry is essential for human identification, but traditional manual techniques are often time-intensive and prone to subjectivity. The integration of artificial intelligence (AI) presents a transformative opportunity to enhance accuracy and efficiency across various forensic odontology applications. This systematic review consolidates and analyzes the existing literature on AI applications in forensic dentistry. Materials and Methods: A comprehensive literature search was conducted in the PubMed and Scopus databases following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Studies published up to August 2025 were considered for inclusion. Results: A total of 116 studies were analyzed. The primary applications included age estimation (54 studies), gender estimation (33 studies), and human/dental identification (26 studies). Bite mark analysis (1 study) and facial reconstruction (2 studies) were identified as underexplored areas. Convolutional neural networks (CNNs) were the most frequently employed AI algorithm, utilized in 64% (74/116) of studies, while orthopantomograms (OPGs) were the predominant imaging modality in 72% (84/116) of investigations. Quantitative results demonstrated significant promise: age estimation models achieved a mean absolute error (MAE) as low as 0.0079 years and a peak classification accuracy of 99.98%; gender estimation models attained accuracies between 68% and 98%; and human identification models reached up to 100% accuracy in optimal conditions, though sensitivity dropped to 69% in complex postmortem cases. Emerging techniques, such as microbiome analysis and generative adversarial networks (GANs), were noted as innovative future directions. Conclusion: AI demonstrates significant potential to improve accuracy and reduce processing time in core forensic dentistry tasks, particularly age and gender estimation. However, challenges related to data privacy, algorithmic bias, and legal admissibility persist. Future research should prioritize the development of explainable AI models, standardized and diverse datasets, and robust regulatory frameworks to ensure ethical and trustworthy integration into forensic practice.

Computer Vision (CV)-based human identification using orthopantomograms (OPGs) has the potential to identify unknown deceased individuals by comparing postmortem OPGs with a comprehensive antemortem CV database. However, the growing size of the CV database leads to longer processing times. This study aims to develop a standardized and reliable Convolutional Neural Network (CNN) for age estimation using OPGs and integrate it into the CV-based human identification process. The CNN was trained on 50,000 OPGs, each labeled with ages ranging from 2 to 89 years. Testing included three postmortem OPGs, 10,779 antemortem OPGs, and an additional set of 70 OPGs within the context of CV-based human identification. Integrating the CNN for age estimation into CV-based human identification process resulted in a substantial reduction of up to 96% in processing time for a CV database containing 105,251 entries. Age estimation accuracy varied between postmortem and antemortem OPGs, with a mean absolute error (MAE) of 2.76 ± 2.67 years and 3.26 ± 3.06 years across all ages, as well as 3.69 ± 3.14 years for an additional 70 OPGs. In conclusion, the incorporation of a CNN for age estimation in the CV-based human identification process significantly reduces processing time while delivering reliable results.

DNA methylation markers in combination with skeletal and dental ages to improve age estimation in children

As a result of the global increase in migration movements in recent years, there is a growing demand for age estimates of live persons. The persons under examination are mostly foreigners without valid identification documents whose genuine age needs to be clarified for legal purposes. In many countries, the age thresholds relevant for criminal, civil, and asylum proceedings lie between 14 and 22 years of age. In line with recommendations issued by the Study Group on Forensic Age Diagnostics, for determining the age of live subjects a forensic age estimate should combine the results of a physical examination, an X-ray of the hand, and a dental examination which records dentition status and evaluates an orthopantomogram. To assess the age of persons who are assumed to be at least 18 years old, an additional roentgenographic or CT examination of the clavicles is recommended. If there is no legal justification for X-ray examinations, the range of possible methods is limited to a physical and a dental examination. The present paper addresses the influence of ethnicity on the examined developmental systems. In so doing, the authors conclude that forensic age estimates should pay due heed to the proband’s socio-economic status and ethnic origin. The effective doses from X-ray-examinations for forensic age estimations are given. There is no fear that the amount of radiation the examined individuals are exposed to during these X-ray examinations will have a detrimental effect on the person's health. At the end, research desiderata will be mentioned.

One of the major tasks in the forensic field is age estimation in living individuals, especially in adolescents and young adults. The X-ray examination of left hand, panoramic radiograph and CT scan of the sternal end of clavicles are mature means that are widely used. However, the X-ray technique has great radiation on the human body, and imaging radiation for non-diagnosis and treatment purposes does not conform to the current mainstream medical ethics. MRI is nonradioactive tomographic imaging and is one of the research and development directions in forensic age estimation in living individuals now. This paper summarizes the common indicators and analysis methods of MRI in previous research of age estimation, in order to get better understanding of its trends and provide a clue for future relevant studies.

Background: Forensic age estimation and its medicolegal implications are gaining lot of importance mainly to determine the age of undocumented youths and to differentiate juvenile from adults. Radiographic evaluation of third molars in terms of its stages of development, its position and extent of mineralization provides an important tool for age estimation, especially in middle teens and early twenties. Hence, this study was designed to assess the reliability of lower third molar development as an indicator for chronological age estimation in young adults of South Indian Population. Material and methods: The study sample comprised of OPG’s (orthopantamographs) of 188 subjects from a restricted geographic area and known chronological age and gender. Radiographs were evaluated using Demerjian’s method to assess the various stages of mandibular right and left third molar development. Data obtained were subjected to statistical analysis. Regression analysis was performed to obtain formula for dental age calculation. Results: Study revealed strong correlation between Demirjians third molar developmental stages and chronological age. Our study also showed that accuracy of age estimation is enhanced when bilateral third molar is evaluated and used for age estimation. The probability of an individual being adult at developmental stage G is still remains uncertain, however at developmental stage H it is almost assured. Conclusion: Several research groups have evaluated the reliability of third molars in age estimation but consensus on the usefulness of these teeth in forensic view has not been reached. The present study has discussed the utility of 3rd molar calcification stage in differentiating between juvenile and adolescent. Demirjian’s stage H can be used as cut off point for determining the majority/minority status of the individual.

Background Forensic age estimation plays a crucial role in medico-legal investigations, immigration cases, and criminal proceedings. Skeletal maturity indicators such as iliac crest ossification have been widely studied to assess chronological age. While conventional radiography has been frequently used for age estimation, variability in ossification timelines across populations necessitates further validation. The iliac crest apophysis follows a predictable maturation pattern, making it a valuable marker for forensic age determination. This study evaluates the reliability of iliac crest ossification staging in forensic age estimation using pelvic X-rays. Objective To assess the ossification status of the iliac crest apophysis in individuals aged 10 to 29 years using Kreitner’s four-stage classification and determine its applicability in forensic age estimation. Methods A retrospective cross-sectional study was conducted at Hayatabad Medical Complex, Peshawar, involving 549 individuals who underwent pelvic X-rays between January 2023 and August 2024. Participants included 277 females (50.5%) and 272 males (49.5%), aged 10–29 years. Digital pelvic radiographs were analyzed by an experienced radiologist blinded to patient age. Iliac crest ossification was classified into four stages: stage 1 (absence of ossification), stage 2 (ossification without apophyseal fusion), stage 3 (partial fusion), and stage 4 (complete fusion). Spearman’s correlation was used to assess the relationship between age and ossification stages, with P &lt; 0.05 considered statistically significant. Results Stage 2 ossification was first observed at 12 years in females and 13 years in males. Stage 3 was first detected at 15 years in females and 17 years in males. Stage 4 ossification was identified at a minimum age of 17 years in males on both sides of the pelvis, while in females, it appeared at 20 years on the right iliac crest. The maximum observed age for stage 2 was 24 years in females and 25 years in males, while stage 3 and 4 ossification extended up to 28 years. A significant correlation was observed between age and iliac crest ossification (P = 0.0001), with correlation coefficients of r = 0.719 (right) and r = 0.716 (left) in males and r = 0.724 (right) and r = 0.700 (left) in females. Conclusion Iliac crest ossification assessed via conventional radiography provides a reliable skeletal marker for forensic age estimation. When used alongside other skeletal maturity indicators, this method enhances the accuracy of age determination. Further research with larger and diverse populations is needed to standardize age estimation criteria for forensic and clinical applications.

Forensic age estimation in living or dead individuals is an important and difficult task. Teeth are considered a reliable indicator of age and provide a number of parameters for age prediction. The aim of this study was to estimate the age in a sample of Egyptian population using the tooth coronal index (TCI) in the mandibular permanent 1st molars in dental panoramic radiographs. Digital panoramic radiographs from 170 participants of known age and sex were included. They were used to measure the crown height (CH) and the coronal pulp cavity height (CPCH) in millimeters. Then, TCI was calculated according to Ikeda et al., (1985). Non significant statistical difference was observed between females and males as regards the TCI. A significant negative linear correlation between chronological age and values of TCI was detected. The correlation coefficients were -0.988, -0.987 and -0.986 in all participants, female participants and male participants respectively. The values of TCI were then subjected to regression analysis to derive equations for age estimation. Regression equations showed coefficient of determination (R2) of 0.984, 0.986 and 0.982 in all participants, female participants and male participants respectively. It was concluded that calculation of the TCI from the mandibular permanent 1st molars is a good predictor of age among Egyptians. Regression equations based on TCI were established and they proved to be very useful for prediction of age with no need for sex specific equations. So, calculation of the TCI is recommended as an easy method for age estimation. Further studies on larger sample size and studies using teeth other than the mandibular permanent 1st molars are needed.

Dental age assessment of young Iranian adults using third molars: A multivariate regression study

Forensic age estimation in living individuals is particularly challenging near legally relevant age thresholds after dental maturation is complete. This study aimed to evaluate the applicability of periodontal ligament (PDL) visibility staging in third molars as a post-maturation marker for threshold-based forensic age estimation, with specific focus on the 21-year age cut-off, in a Turkish population. A retrospective analysis was conducted on 1,049 digital orthopantomograms (OPGs) of individuals aged 15–25 years. PDL visibility in maxillary and mandibular third molars was assessed according to the Olze et al. staging system. Associations between PDL stages, chronological age, sex, and tooth position were analyzed. Diagnostic performance for identifying individuals aged ≥ 21 years was evaluated using receiver operating characteristic (ROC) analysis. Advanced PDL visibility stages were significantly associated with increasing age. Mandibular third molars provided the most informative staging patterns. ROC analysis demonstrated good discriminative accuracy for the 21-year threshold, with an area under the curve (AUC) of 0.79, indicating balanced sensitivity and specificity for threshold-based classification. Substantial intra- and inter-observer agreement confirmed methodological reliability. PDL visibility staging in third molars represents a reliable and reproducible complementary tool for forensic age estimation around the 21-year threshold in a Turkish population. Rather than exact age prediction, this method supports probabilistic decision-making in legally relevant age assessment and may contribute to a multimodal forensic age estimation framework.

Objective To evaluate the accuracy of different convolutional neural networks (CNN),representative deep learning models,in the differential diagnosis of ameloblastoma and odontogenic keratocyst,and subsequently compare the diagnosis results between models and oral radiologists. Methods A total of 1000 digital panoramic radiographs were retrospectively collected from the patients with ameloblastoma (500 radiographs) or odontogenic keratocyst (500 radiographs) in the Department of Oral and Maxillofacial Radiology,Peking University School of Stomatology.Eight CNN including ResNet (18,50,101),VGG (16,19),and EfficientNet (b1,b3,b5) were selected to distinguish ameloblastoma from odontogenic keratocyst.Transfer learning was employed to train 800 panoramic radiographs in the training set through 5-fold cross validation,and 200 panoramic radiographs in the test set were used for differential diagnosis.Chi square test was performed for comparing the performance among different CNN.Furthermore,7 oral radiologists (including 2 seniors and 5 juniors) made a diagnosis on the 200 panoramic radiographs in the test set,and the diagnosis results were compared between CNN and oral radiologists. Results The eight neural network models showed the diagnostic accuracy ranging from 82.50% to 87.50%,of which EfficientNet b1 had the highest accuracy of 87.50%.There was no significant difference in the diagnostic accuracy among the CNN models (P=0.998,P=0.905).The average diagnostic accuracy of oral radiologists was (70.30±5.48)%,and there was no statistical difference in the accuracy between senior and junior oral radiologists (P=0.883).The diagnostic accuracy of CNN models was higher than that of oral radiologists (P<0.001). Conclusion Deep learning CNN can realize accurate differential diagnosis between ameloblastoma and odontogenic keratocyst with panoramic radiographs,with higher diagnostic accuracy than oral radiologists.

Above or below 14years? An orthopantomographic study based on chronological course of eruption of mandibular premolars and second molars in a sample of south Indian children.

Leaf disease detection is a crucial task in modern agriculture, aiding in early diagnosis and prevention of crop infections. In this research paper, authors present a comprehensive study comparing nine widely used pre-trained models, namely DenseNet201, EfficientNetB3, EfficientNetB4, InceptionResNetV2, MobileNetV2, ResNet50, ResNet152, VGG16, and Xception, with our newly developed custom CNN (Convolutional Neural Network) for leaf disease detection. The objective is to determine if our custom CNN can match the performance of these established pre-trained models while maintaining superior efficiency. The authors trained and fine-tuned each pre-trained model and our custom CNN on a large dataset of labeled leaf images, covering various diseases and healthy states. Subsequently, the authors evaluated the models using standard metrics, including accuracy, precision, recall, and F1-score, to gauge their overall performance. Additionally, the authors analyzed computational efficiency regarding training time and memory consumption. Surprisingly, our results indicate that the custom CNN performs comparable to the pre-trained models, despite their sophisticated architectures and extensive pre-training on massive datasets. Moreover, our custom CNN demonstrates superior efficiency, outperforming the pre-trained models regarding training speed and memory requirements. These findings highlight the potential of custom CNN architectures for leaf disease detection tasks, offering a compelling alternative to the commonly used pre-trained models. The efficiency gains achieved by our custom CNN can be beneficial in resource-constrained environments, enabling faster inference and deployment of leaf disease detection systems. Overall, our research contributes to the advancement of agricultural technology by presenting a robust and efficient solution for the early detection of leaf diseases, thereby aiding in crop protection and yield enhancement.