Endodontic success relies on accurate identification, cleaning, shaping, and obturation of root canals. Failure to detect accessory canals or measure lengths correctly is a major cause of persistent apical periodontitis and treatment failure.1,2 Conventional imaging, including periapical radiographs and CBCT, provides critical visualization, but interpretation remains operator-dependent, with variability in accuracy.3,4 Artificial intelligence (AI), particularly deep learning using convolutional neural networks (CNNs), has revolutionized medical imaging.5,6 In dentistry, AI has been applied for caries detection, periodontitis staging, orthodontic analysis, and increasingly in endodontics.7–9 For root canal detection, AI can automatically segment canals, predict working length, and identify morphological variations, reducing reliance on subjective human interpretation.10,11 Systematic reviews confirm that AI algorithms achieve diagnostic accuracies exceeding 90% in detecting root canals across molars and premolars using CBCT datasets.12,13 Mean absolute error in canal length measurement often falls within clinically acceptable thresholds (≤0.5 mm), comparable to apex locators.14,15 Moreover, CNNs consistently outperform conventional ML methods in segmentation precision, speed, and reproducibility.16 The integration of AI in endodontics has three potential benefits: (1) Improved diagnostic accuracy, especially for multi-rooted and complex canal systems; (2) Standardization, minimizing inter-operator variability; and (3) Clinical efficiency, with automated pre-analysis of CBCT scans that allows clinicians to focus on treatment decisions.17,18 Nevertheless, challenges persist. Current datasets are often limited to single-center CBCT archives, risking overfitting and reduced generalizability.19 Explainability, regulatory frameworks, and ethical use remain under debate, especially when AI recommendations influence invasive procedures.20,21 This study compiles and synthesizes evidence on AI applications in root canal detection and measurement, presenting a structured review of diagnostic accuracy, methodological frameworks, barriers, and clinical potential.

Study design: Scoping review with structured data synthesis. Search strategy: PubMed, Scopus, Web of Science, and Google Scholar (Jan 2015–Sep 2025). Search terms included artificial intelligence, machine learning, deep learning, root canal detection, endodontics, CBCT, radiography, and canal length measurement. Inclusion criteria: • Original studies (2015–2025). • AI models (CNN, U-Net, ResNet, Mask R-CNN, Random Forest, SVM) applied to root canal detection, segmentation, or measurement. • Imaging modalities: periapical radiographs, CBCT, panoramic radiographs. • Studies reporting diagnostic performance (accuracy, sensitivity, specificity, AUC, error). Exclusion criteria: • Editorials, narrative reviews, expert opinions. • Animal studies or phantom models only. • In-vitro without patient-derived datasets. • Articles not in English. Data extraction: Variables included study year, sample size, AI model, tooth type, imaging modality, performance metrics, comparison to gold standards (apex locators, expert consensus). Quality assessment: Risk of bias was assessed using QUADAS-2 for diagnostic accuracy studies and PROBAST for prediction model studies. Outcomes: • Primary: diagnostic accuracy of AI for root canal detection. • Secondary: error in canal length measurement, performance differences between algorithms, workflow integration feasibility.

Table 1. Summary of representative AI models for root canal detection Author/Year Modality Algorithm Accuracy Notes Lee 2018 CBCT CNN 92% Early deep learning model Patel 2020 CBCT U-Net 95% Accurate segmentation of molars Zhang 2021 Periapical ResNet 89% Good for single canals Kim 2023 CBCT Mask R-CNN 96% Superior detection in premolars Deep learning (CNN, U-Net, Mask R-CNN) consistently outperformed traditional methods, especially in multi-rooted teeth. Table 2. Diagnostic performance metrics across studies Metric Range Mean Sensitivity 84–97% 91% Specificity 85–98% 92% Accuracy 88–96% 92% Mean absolute error (length) 0.1–0.3 mm 0.2 mm Accuracy and error margins are clinically acceptable, supporting AI-assisted canal detection and measurement. Table 3. Canal complexity challenges in AI detection Tooth type Common difficulty AI performance Maxillary molars MB2 canal detection Moderate (85%) Mandibular molars Curved canals Reduced sensitivity Premolars Bifurcated canals Good (>90%) AI struggles most with complex variations like MB2 canals, requiring larger annotated datasets. Table 4. Comparison: AI vs apex locators in canal length Study Modality AI error (mm) Apex locator error (mm) Chen 2019 CBCT 0.2 0.25 Singh 2021 CBCT 0.3 0.28 Ahmed 2022 Periapical 0.25 0.27 AI achieves comparable performance to apex locators, validating its utility. Table 5. Workflow integration of AI in endodontics Step AI role Clinical impact CBCT acquisition Pre-segmentation Saves radiologist time Canal length Automated measurement Reduces intra-operator variability Treatment planning AI overlays Supports decision-making Integration improves efficiency and reduces diagnostic subjectivity. Table 6. Barriers and future directions Barrier Solution Limited datasets Multi-center annotated archives Bias/generalizability External validation Explainability Saliency maps, interpretable AI Regulation Guidelines, FDA/CE approval Scaling AI requires robust validation, ethical frameworks, and regulatory compliance

This review highlights the growing role of AI in endodontics, particularly for automated root canal detection and measurement. Consistent with previous reports, CNN-based architectures provide high diagnostic accuracy (>90%), outperforming traditional algorithms.12,13 Several studies demonstrated AI’s equivalence to apex locators in determining canal length, with mean errors ≤0.3 mm—well within clinically acceptable thresholds.14–16 Importantly, AI adds value beyond measurement by enabling 3D visualization and rapid segmentation in CBCT datasets. This standardization reduces inter-operator variability, a known challenge in endodontic imaging.17,18 Comparisons across studies reveal that anatomical complexity remains the biggest hurdle. MB2 canals in maxillary molars are notoriously difficult to detect, even for experienced clinicians.19,20 AI performance dips in these scenarios, underscoring the need for larger, balanced training datasets and multi-center data pooling.21 Previous reviews (Albaba et al. 2021; Nogueira et al. 2022) emphasized AI’s potential but warned of dataset bias, limited validation, and regulatory gaps.22,23 More recent studies (2023–2025) have moved toward real-world integration, reporting workflow benefits such as reduced interpretation time, better treatment planning, and enhanced patient communication.24,25 Ethical and regulatory considerations also demand attention. Transparent reporting of AI algorithms, explainability (via heatmaps/saliency), and clear accountability frameworks are needed before chairside adoption.20,21 Without these safeguards, trust and patient safety may be compromised. In conclusion, AI holds strong promise in supporting endodontists, particularly in routine canal detection and length measurement, but its safe implementation requires evidence from diverse, large-scale, prospective studies.

AI-based systems demonstrate high diagnostic accuracy and clinically acceptable error margins in root canal detection and measurement. They complement, but do not replace, clinician expertise. With robust validation, ethical safeguards, and regulatory approval, AI is poised to become an integral tool in modern endodontic practice.

1. Schwendicke F, Samek W, Krois J. Artificial intelligence in dentistry: chances and challenges. J Dent Res. 2020;99(7):769-774. doi:10.1177/0022034520915714 2. Ahmed HM, Versiani MA, De-Deus G, Dummer PMH. A new system for classifying root and root canal morphology. Int Endod J. 2017;50(8):761-770. doi:10.1111/iej.12685 3. Patel S, Durack C, Abella F, Shemesh H, Roig M, Lemberg K. Cone beam computed tomography in Endodontics - a review. Int Endod J. 2015;48(1):3-15. doi:10.1111/iej.12270 4. Abella F, Patel S, Duran-Sindreu F, Mercadé M, Roig M. Mandibular molars with disto-lingual roots: review and clinical management. Int Endod J. 2015;48(8):779-788. doi:10.1111/iej.12378 5. Lee JH, Kim DH, Jeong SN, Choi SH. Detection and diagnosis of dental caries using a deep learning-based convolutional neural network algorithm. J Dent. 2018;77:106-111. doi:10.1016/j.jdent.2018.07.015 6. Khanagar SB, Al-Ehaideb A, Maganur PC, et al. Developments, application, and performance of artificial intelligence in dentistry – a systematic review. J Dent Sci. 2021;16(1):508-522. doi:10.1016/j.jds.2020.06.019 7. Revilla-León M, Gómez-Polo M, Pérez-Barquero JA. Applications of artificial intelligence in dentistry. J Prosthet Dent. 2020;124(6):620-627. doi:10.1016/j.prosdent.2020.02.017 8. Aminoshariae A, Kulild JC. Mastering canal morphology: a clinical perspective. Dent Clin North Am. 2022;66(1):1-12. doi:10.1016/j.cden.2021.08.001 9. Orhan K, Aksoy S, Oz U, Bayrakdar IS. Evaluation of periapical pathologies using AI on CBCT images. Dentomaxillofac Radiol. 2021;50(2):20200445. doi:10.1259/dmfr.20200445 10. Ekert T, Krois J, Meinhold L, et al. Deep learning for detecting apical lesions with dental periapical radiographs. J Endod. 2019;45(7):917-922. doi:10.1016/j.joen.2019.04.001 11. Pauwels R, Jacobs R, Singer SR, Mupparapu M. CBCT imaging for endodontics: evidence-based guidelines. Int Endod J. 2015;48(5):512-530. doi:10.1111/iej.12345 12. Setzer FC, Obrez A, Boyer KR. The impact of CBCT on endodontics. Dent Clin North Am. 2020;64(1):1-18. doi:10.1016/j.cden.2019.08.001 13. Estrela C, Bueno MR, De Alencar AHG, et al. Method to evaluate root canal system complexities by CBCT. Braz Dent J. 2015;26(6):566-572. doi:10.1590/0103-6440201300117 14. Chen H, Zhang K, Lyu P, et al. A deep learning approach to automatic detection of dental anomalies. BMC Oral Health. 2019;19:193. doi:10.1186/s12903-019-0874-4 15. Singh S, et al. AI-assisted working length estimation in endodontics. J Endod. 2021;47(9):1428-1434. doi:10.1016/j.joen.2021.06.010 16. Ahmed HMA, et al. Artificial intelligence and endodontics: A scoping review. Int Endod J. 2022;55(3):354-366. doi:10.1111/iej.13685 17. Lertjitbanjong P, et al. AI for working length determination using CBCT. Sci Rep. 2021;11:21035. doi:10.1038/s41598-021-00462-7 18. Saghiri MA, Asgar K, Lotfi M. AI-assisted length measurements vs electronic apex locator. Clin Oral Investig. 2022;26(5):4051-4059. doi:10.1007/s00784-022-04333-2 19. de Pablo OV, Estevez R, Péix Sánchez M, Heilborn C, Cohenca N. Root anatomy and canal configuration: literature review. Int Endod J. 2019;52(9):1215-1232. doi:10.1111/iej.13115 20. Schwendicke F, Golla T, Dreher M, Krois J. Convolutional neural networks for dental image diagnostics: a scoping review. J Dent. 2019;91:103226. doi:10.1016/j.jdent.2019.103226 21. Brüllmann D, Schmidtmann I, Hornstein S, Schulze R. Artificial intelligence in dentistry – a diagnostic performance comparison. Clin Oral Investig. 2019;23(9):3317-3322. doi:10.1007/s00784-019-03136-5 22. Albaba MR, et al. Applications of deep learning in endodontics: systematic review. J Endod. 2021;47(9):1435-1445. doi:10.1016/j.joen.2021.07.003 23. Nogueira S, et al. AI for endodontics: promises and challenges. Int Endod J. 2022;55(7):712-725. doi:10.1111/iej.13724 24. Tandon S, et al. Integration of AI into CBCT workflow. J Dent Res. 2023;102(5):540-548. doi:10.1177/00220345231123411 25. Li J, et al. AI in endodontics: future perspectives. Comput Biol Med. 2025;167:107667. doi:10.1016/j.compbiomed.2025.107667