AI systems for caries detection/segmentation on dental X-ray images (predominantly convolutional neural networks; also ANNs, DNNs, DCNN, FCNN, SVMs, decision trees, random forests; e.g. EfficientNet, ResNet, AlexNet) vs Human dentists / trained dental professionals or traditional radiographic interpretation (reference standard / reference models): Sensitivity
Pooling 9 studies (17,190 participants), AI sensitivity for caries detection showed no significant association (pooled OR 1.258, 95% CI 0.493-2.540) with high heterogeneity (I2=86%); ungraded.
What each review found
Include earlier review versions| Finding | Effect | Studies | GRADE certainty |
|---|---|---|---|
AI systems for caries detection/segmentation on dental X-ray images (predominantly convolutional neural networks; also ANNs, DNNs, DCNN, FCNN, SVMs, decision trees, random forests; e.g. EfficientNet, ResNet, AlexNet) vs Human dentists / trained dental professionals or traditional radiographic interpretation (reference standard / reference models) Sensitivity · Patients undergoing dental X-ray (radiographic) imaging for caries detection · Dental radiography (digital periapical, bitewing, panoramic; OCT/micro-CT in one study); studies 2017-2023 across China, South Korea, Brazil, Turkey, France, USA, UK, Taiwan, Germany, India, Netherlands, Overall (all studies pooled) Pooling 9 studies (17,190 participants), AI sensitivity for caries detection showed no significant association (pooled OR 1.258, 95% CI 0.493-2.540) with high heterogeneity (I2=86%); ungraded. Accuracy of artificial intelligence in caries detection: a systematic review and meta-analysis. (2025) All outcomes | OR 1.258 [0.493, 2.540] | 9 studies · 17190 participants | GRADE not reported |
Findings are shown side by side, never pooled across reviews.
The reviews
| Review | Searched up to | Included studies | AMSTAR 2 |
|---|---|---|---|
| Accuracy of artificial intelligence in caries detection: a systematic review and meta-analysis. 2025 All outcomes | Up to January 2024 (data collected up to January 2024; electronic search completed 19 May 2023, final search included research up to January 2024) | 35 | Not applicable |
Studies behind the findings
The primary studies the full-text reviews tie to this outcome. Reviews with headline outcomes only do not list them here.
- Reduction-induced surface modification of human hair. · 2010 · included by 1 of these review
- Devlin 2021 · included by 1 of these review
- Mertens 2021 · included by 1 of these review
How reviews are matched to this question
Reviews word the same question differently. Four decision models (TypeSafe's Jev, Liquid AI's d1-ultra, SubQ Snap and Baseten's Elephant) judged each wording of the population, intervention, comparator and outcome against the others; two wordings count as the same when at least three of them agree and none clearly disagrees, and findings count as this question only when all four elements match. Timing, setting and subgroup never split a question: they are shown on each finding.