Background: Root canal treatment (RCT) remains the most common clinical intervention in dentistry, treating a wide range of pulp-related conditions and structural tooth injuries. RCT begins by partially removing the crown to access and remove the infected pulp through a meticulous process of cleaning, reshaping, and irrigating the canal. This prepared canal is filled with gutta-percha as the endodontic filling material (EFM), and adhesive cement is used to seal the canal before final restoration with a crown. The success of RCT, however, hinges on accurately measuring and applying the EFM volume within the canal. Current measurement practices using 2D radiographic imaging fall short in accuracy due to operator limitations and the inherent lack of depth and clarity in 2D images, which contribute to high variability in EFM application. CBCT has emerged as a viable alternative, offering comprehensive 3D imaging for a more accurate EFM volume assessment, yet its high radiation levels, cost, and accessibility issues pose challenges. Thus, with increasing demand for safer and cost-effective alternatives, AI-driven enhancements in 2D imaging are being explored to achieve CBCT-level accuracy without its drawbacks, marking a significant step forward in RCT reliability and success rates. Purpose: To explore the potential of GANs and Image–J integration on 2D images for estimation of EFM measurement. Method: A literature review of studies published within the past five years from sources like ResearchGate, ScienceDirect, PubMed, and Google Scholar. Result: This study explores the integration of artificial intelligence, specifically convolutional neural networks (CNN) and generative adversarial networks (GANs), to enhance the accuracy of endodontic filling material (EFM) measurements in root canal treatments (RCT). CNN's structure, comprising multilayered networks with pooling layers, has shown potential in medical imaging by segmenting anatomical details for EFM measurement. However, challenges remain due to CNN's reliance on extensive datasets and its susceptibility to misinterpretation when faced with complex anatomical variations. GANs, combining a generator and discriminator, address these limitations by enabling unsupervised learning, allowing the generation of photorealistic data samples for more precise EFM measurement. This bidirectional learning system refines measurement accuracy through continuous feedback between the generator and discriminator, achieving superior outcomes across parameters like radiopacity, dimensional consistency, classification performance, and sealing ability. Furthermore, the AI model is enhanced by integrating Image-J, a Java-based image analysis software with precise 2D measurement capabilities. Despite its effectiveness in analyzing length and volume, Image-J has limitations due to its dependence on the operator’s interpretation, especially with low-quality 2D images. The combined use of GANs with Image-J allows for automated, detailed border detection of the pulp anatomy, improving measurement accuracy even with complex root canal forms. The results suggest that this integrated AI approach can advance clinical practices by providing precise, rapid, and consistent EFM measurements, potentially overcoming the limitations of traditional 2D radiographic methods and addressing CBCT’s drawbacks in radiation exposure and accessibility. Conclusion: The integration of generative adversarial networks (GANs) with Image-J offers a promising advancement in the measurement of endodontic filling materials (EFM) for root canal treatments based on 2D imaging. GANs enhance Image-J’s measurement precision and accuracy by utilizing bidirectional deep learning, which enables continuous learning and refinement without requiring extensive external inputs. This approach not only improves EFM estimation but also presents a viable alternative to cone-beam computed tomography (CBCT), reducing potential health risks and economic burdens associated with CBCT. The combined GANs and Image-J system thus supports a more accessible, efficient, and precise methodology for EFM assessment in clinical dentistry.
Read full abstract