| Title |
Artificial Intelligence Enhances Diagnostic Accuracy in Endodontic Treatment |
| Clinical Question |
In patients undergoing endodontic therapy, does the use of artificial intelligence, compared to traditional clinical techniques, improve diagnostic accuracy and treatment planning? |
| Clinical Bottom Line |
For patients undergoing endodontic therapy, the use of artificial intelligence systems can increase diagnostic accuracy in detecting periapical lesions, root morphologies, and vertical root fractures. Across three systematic reviews, AI models showed sensitivity and specificity that met or exceeded that of clinical judgment alone. Limitations included small sample size and variability in models used. With standardized protocols and larger-scale studies, AI models could significantly enhance diagnostic precision and efficiency in endodontic treatment planning. |
| Best Evidence |
|
| PubMed ID |
Author / Year |
Patient Group |
Study type
(level of evidence) |
| 36766519 | Khanagar/2021 | Patients receiving endodontic diagnoses across 37 original studies; 21 studies using CNNs | Systematic Review | | Key results | Convolutional neural networks (CNNs), an AI diagnostic model used in endodontics, demonstrated strong accuracy with sensitivity and specificity above 90% when used to detect periapical lesions, canal morphologies, and vertical root fractures. In addition, low risk of selection bias was reported (90%) for patient selection. Most of the studies included in this review were limited by data coming from a single clinic and single radiographic machine. This would decrease generalizability due to moderate applicability (70%). AI was recommended as a supplemental diagnostic tool to support clinical decision-making and expedite treatment planning. | | 38851523 | Pul/2024 | Patients receiving radiographs in 24 studies | Systematic Review and Meta-Analysis | | Key results | This systematic review and meta-analysis concluded that AI detection of periapical radiolucencies provided results with a sensitivity of 92% (95% CI: 89%-95%) and specificity of 90% (95% CI: 87%-92%). Risk of bias was considered unclear or high according to QUADAS-2, a tool for quality assessment of diagnostic accuracy studies, which took into consideration patient selection, index test, reference standard, and flow and timing. A limited number of eligible studies with relatively small sample sizes highlights the need for further research into the application of AI as a tool. Overall, the study emphasized AI’s potential, but called for further research. | | 36548872 | Ramezanzade/2023 | 24 Studies | Systematic Review | | Key results | This systematic review analyzed the effectiveness of AI detection of radiographic features such as periapical lesions, root and canal morphology, vertical root fractures, apical foramen location, and retreatment prediction with >85% accuracy. AI demonstrated high diagnostic potential in endodontics. However, 58% of included studies exhibited methodological bias, reducing reliability and limiting generalizability, despite promising accuracy metrics. This underscores the need for more rigorous, standardized research to validate the clinical application of AI in routine practice of endodontic diagnosis. | |
| Evidence Search |
“artificial intelligence” AND (“endodontics” OR “periapical radiolucency” OR “root canal”) |
Comments on
The Evidence |
In Khanagar et al., QUADAS-2 was used to assess risk of bias, GRADE was used to weigh strength of evidence, and AMSTAR-2 was used to evaluate pre-registration protocol and data on extraction process. Although patient selection bias was low, insufficient methodological detail reduced reliability. Pul and Schwendicke also used QUADAS-2. While this study demonstrated unclear to high risk of bias, sensitivity and specificity were considered high. This may be due to the model being tested under ideal rather than real world clinical conditions. Lastly, Ramezanzade et al. followed PRISMA guidelines and used QUADAS-2 to assess risk of bias in included studies. Because high heterogeneity and methodological bias was found among studies used, generalizability of the results is limited and validity is weakened. |
| Applicability |
With the growing interest of AI use in dentistry, it is important to analyze the role it plays in maximizing efficiency and productivity in future practice. Although current literature supports AI’s strong performance in endodontic radiographic interpretation, further research and protocol standardization are needed before widespread clinical adoption. Each study suggested promising benefits in future treatment planning with growing accessibility. Therefore, it is suggested that AI be used as a tool to improve endodontic diagnostic accuracy but verified through clinical judgment. |
| Specialty |
(Endodontics) (General Dentistry) |
| Keywords |
root canal treatment, RCT, endodontics, artificial intelligence, AI
|
| ID# |
3581 |
| Date of submission |
06/24/2025 |
| E-mail |
lexi.tan21@yahoo.com |
| Author |
Alexandria I. Tan |
| Co-author(s) |
Sophia Hung |
| Co-author(s) e-mail |
sophie.hung3@gmail.com |
| Faculty mentor |
Kelly C. Lemke, DDS, MS |
| Faculty mentor e-mail |
lemkek@uthscsa.edu |
| |
|
Basic Science Rationale
(Mechanisms that may account for and/or explain the clinical question, i.e. is the answer to the clinical question consistent with basic biological, physical and/or behavioral science principles, laws and research?) |
| None available | |
 |
Comments and Evidence-Based Updates on the CAT
(FOR PRACTICING DENTISTS', FACULTY, RESIDENTS and/or STUDENTS COMMENTS ON PUBLISHED CATs) |
| None available | |