Application of Artificial Intelligence in Digital Radiographs for Early Detection of Carious Lesions

Abstract

Introduction: Dental caries is one of the most prevalent chronic diseases and a relevant
public health issue. Early detection is essential to prevent complications and reduce invasive
treatments. In this context, artificial intelligence (AI), especially through machine and deep
learning algorithms, has shown promise in the analysis of digital radiographs, increasing
diagnostic accuracy and standardizing results. Objectives: To evaluate, through an
integrative review, the effectiveness of AI use in digital radiographs for the early detection of
carious lesions, comparing it with conventional diagnosis and highlighting benefits and
limitations for dental practice. Methodology: An integrative review was carried out in the
MEDLINE (PubMed), BVS, and SciELO databases using the descriptors “artificial intelligence,”
“dental radiography,” “dental caries,” “early detection,” “machine learning,” and “deep
learning.” Studies published between 2021 and 2025 addressing AI in digital radiographs
were included; incomplete texts, theses, proceedings, and studies outside the specified period
were excluded. The selection followed the PICO strategy: patients evaluated by digital
radiographs (P); application of AI for caries detection (I); comparison with conventional
assessment (C); diagnostic accuracy and clinical benefits (O). Data were synthesized
qualitatively due to methodological heterogeneity. Conclusion: AI applied to radiographic
analysis demonstrated high accuracy in the early detection of caries, often equal to or
greater than human evaluation. Systems based on neural networks and deep learning offer
speed, diagnostic standardization, and clinical decision support, favoring preventive dentistry.
Despite limitations such as costs, training requirements, and dependence on image quality, AI
is consolidated as a complementary tool to optimize clinical workflows, reduce diagnostic
errors, and improve dental care.