Diagnostic accuracy of artificial intelligence in detecting periapical lesions: A comparative study with cone-beam computed tomography and conventional periapical radiography.
- 1Faculty of Oral and Dental Medicine, Future University in Egypt, Cairo, Egypt.
- 2Faculty of Dentistry, Cairo University, Cairo, Egypt.
Abstract
Periapical lesions, particularly in their early stages, can be challenging to identify clinically and radiographically, as they may be asymptomatic or obscured by anatomical structures on conventional two-dimensional periapical radiography (2D PR). The diagnosis of periapical pathology may also be subjective and influenced by the examiner's experience, image quality, and interpretation variability. With recent technological advancements, artificial intelligence (AI)-based diagnostic tools have been introduced to support radiographic interpretation and improve the consistency of lesion detection. This study aimed to evaluate the diagnostic accuracy of AI in detecting periapical lesions by comparing its performance with 2D PR and cone-beam computed tomography (CBCT), using CBCT as the reference standard. This diagnostic accuracy study included 177 teeth from adult patients indicated for endodontic treatment. All teeth were examined using conventional 2D PR, CBCT, and AI-based image interpretation. The presence or absence of periapical pathology was recorded for each modality, and CBCT was considered the reference standard. Diagnostic performance was assessed using sensitivity, specificity, positive predictive value, negative predictive value, overall accuracy, receiver operating characteristic (ROC) curve analysis, and agreement testing. Compared with CBCT, AI showed a sensitivity of 84.16%, specificity of 93.42%, and accuracy of 88.14%. Conventional 2D PR showed lower sensitivity (64.36%) despite high specificity (100%). When periapical radiography was used as reference, AI demonstrated higher sensitivity (89.23%) but lower specificity (71.43%). CBCT remained the most reliable reference modality for detecting periapical lesions. AI demonstrated promising diagnostic accuracy and may support radiographic interpretation in endodontic practice.
Related
- Evaluation of Tooth Shade Selection With Conventional Methods and Artificial Intelligence-Integrated Methods: A Comparative Study
- Advancing Biosensing Through Artificial Intelligence: Overcoming Conventional Limitations Towards Intelligent Diagnostic Systems.
- CT-Referenced Diagnostic Performance and Complementary Error Patterns of Artificial Intelligence Versus a Single Emergency Physician in Chest Radiograph Interpretation: A Retrospective Paired Diagnostic Accuracy Study
- Human–artificial intelligence collaboration in emergency medical services: from diagnostic algorithms to shared decision support
- Beyond diagnostic accuracy: navigating the artificial intelligence data tsunami toward value-based implementation in gastroenterology
BibTeX
@article{albochi2026diagnostic,
title = {{Diagnostic accuracy of artificial intelligence in detecting periapical lesions: A comparative study with cone-beam computed tomography and conventional periapical radiography.}},
author = {Mays Maher Albochi and Wael Hussein Kamel and Nihal Ezzat Sabet and Marwa Wagih ElBoraey},
journal = {Journal of clinical and experimental dentistry},
year = {2026},
month = sep,
doi = {10.4317/jced.64327},
url = {https://europepmc.org/article/MED/42999816},
note = {CC BY},
}
RIS
TY - JOUR
TI - Diagnostic accuracy of artificial intelligence in detecting periapical lesions: A comparative study with cone-beam computed tomography and conventional periapical radiography.
AU - Mays Maher Albochi
AU - Wael Hussein Kamel
AU - Nihal Ezzat Sabet
AU - Marwa Wagih ElBoraey
PY - 2026
DA - 2026/09/15
JO - Journal of clinical and experimental dentistry
DO - 10.4317/jced.64327
UR - https://europepmc.org/article/MED/42999816
AB - Periapical lesions, particularly in their early stages, can be challenging to identify clinically and radiographically, as they may be asymptomatic or obscured by anatomical structures on conventional two-dimensional periapical radiography (2D PR). The diagnosis of periapical pathology may also be subjective and influenced by the examiner's experience, image quality, and interpretation variability. With recent technological advancements, artificial intelligence (AI)-based diagnostic tools have been introduced to support radiographic interpretation and improve the consistency of lesion detection. This study aimed to evaluate the diagnostic accuracy of AI in detecting periapical lesions by comparing its performance with 2D PR and cone-beam computed tomography (CBCT), using CBCT as the reference standard. This diagnostic accuracy study included 177 teeth from adult patients indicated for endodontic treatment. All teeth were examined using conventional 2D PR, CBCT, and AI-based image interpretation. The presence or absence of periapical pathology was recorded for each modality, and CBCT was considered the reference standard. Diagnostic performance was assessed using sensitivity, specificity, positive predictive value, negative predictive value, overall accuracy, receiver operating characteristic (ROC) curve analysis, and agreement testing. Compared with CBCT, AI showed a sensitivity of 84.16%, specificity of 93.42%, and accuracy of 88.14%. Conventional 2D PR showed lower sensitivity (64.36%) despite high specificity (100%). When periapical radiography was used as reference, AI demonstrated higher sensitivity (89.23%) but lower specificity (71.43%). CBCT remained the most reliable reference modality for detecting periapical lesions. AI demonstrated promising diagnostic accuracy and may support radiographic interpretation in endodontic practice.
SN - 1989-5488
ER -