Artificial Intelligence in Age Estimation

Abstract

This chapter delves into the pressing need for efficient age estimation methods, particularly in the context of undocumented migrants and forensic investigations. Traditional methods, such as Demirjian’s technique, have been pivotal but are hindered by manual processes and subjectivity. Consequently, the article explores the integration of artificial intelligence (AI) to streamline age estimation processes, citing its potential to automate tasks and enhance accuracy. Various AI models, particularly convolutional neural networks (CNNs), are discussed for their efficacy in dental age estimation, leveraging advancements in deep learning. The article highlights seminal CNN architectures like AlexNet, VGG, Inception, and ResNet, illustrating their evolution and impact on image recognition tasks. Moreover, studies employing CNNs for dental age estimation demonstrate promising results, showcasing the feasibility of automated approaches. The review also encompasses diverse methodologies in dental image processing, including semi-automatic and fully automated segmentation techniques. While previous studies have primarily focused on tooth development stages, recent advancements explore novel approaches like volumetric measurements and object segmentation for improved accuracy. Furthermore, the article underscores the significance of transfer learning and model customization in optimizing CNN performance for age estimation. It discusses the development of customized CNN models from scratch, showcasing their potential to rival conventional methods in accuracy and efficiency. In conclusion, the continued exploration of AI-driven approaches in dental age estimation, emphasizing the need for interdisciplinary collaboration and the integration of advanced imaging modalities, is greatly recommended. Future research directions include extending CNN methodologies with population-specific atlases like the London Atlas, presenting promising avenues for enhancing age estimation accuracy and applicability.