Non-Cochrane · DTA · PM41828006
Deep Learning-Based Dental Caries Diagnosis: A Modality-Stratified Systematic Review and Meta-Analysis of Faster R-CNN and Mask R-CNN.
Lam Quang Tuan, Le Minh Huu Nhat, Fan Fang-Yu, et al. · Diagnostics (Basel, Switzerland) · 2026 · PubMed · doi:10.3390/diagnostics16050731
Findings
Population Dental images (radiographs and intraoral photographs) assessed for dental caries
28 findings
| Faster R-CNN | Reference standard caries status (expert annotation or clinical examination) | AUC for caries detection (HSROC) All images | AUC 0.84 GRADE not reported 11 studies |
| Faster R-CNN | Reference standard caries status (expert annotation or clinical examination) | Specificity for caries detection All images | Specificity 0.814 [0.748, 0.866] GRADE not reported 11 studies |
| Faster R-CNN | Reference standard caries status (expert annotation or clinical examination) | Sensitivity for caries detection All images | Sensitivity 0.717 [0.62, 0.797] GRADE not reported 11 studies |
| Mask R-CNN | Reference standard caries status (expert annotation or clinical examination) | Sensitivity for caries detection All images | Sensitivity 0.856 [0.755, 0.92] GRADE not reported 6 studies |
| Mask R-CNN | Reference standard caries status (expert annotation or clinical examination) | AUC for caries detection (HSROC) All images | AUC 0.95 GRADE not reported 6 studies |
| Mask R-CNN | Reference standard caries status (expert annotation or clinical examination) | Specificity for caries detection All images | Specificity 0.942 [0.879, 0.973] GRADE not reported 6 studies |
| Faster R-CNN | Reference standard caries status (expert annotation or clinical examination) | Sensitivity for caries detection Radiographic images | Sensitivity 0.672 [0.513, 0.799] GRADE not reported 6 studies |
| Faster R-CNN | Reference standard caries status (expert annotation or clinical examination) | Specificity for caries detection Radiographic images | Specificity 0.85 [0.785, 0.898] GRADE not reported 6 studies |
| Faster R-CNN | Reference standard caries status (expert annotation or clinical examination) | AUC for caries detection (HSROC) Radiographic images | AUC 0.86 GRADE not reported 6 studies |
| Mask R-CNN | Reference standard caries status (expert annotation or clinical examination) | Sensitivity for caries detection Radiographic images | Sensitivity 0.863 [0.684, 0.949] GRADE not reported 4 studies |
| Mask R-CNN | Reference standard caries status (expert annotation or clinical examination) | AUC for caries detection (HSROC) Radiographic images | AUC 0.97 GRADE not reported 4 studies |
| Mask R-CNN | Reference standard caries status (expert annotation or clinical examination) | Specificity for caries detection Radiographic images | Specificity 0.965 [0.908, 0.987] GRADE not reported 4 studies |
| Faster R-CNN | Reference standard caries status (expert annotation or clinical examination) | Sensitivity for caries detection Photographic images | Sensitivity 0.773 [0.669, 0.852] GRADE not reported 5 studies |
| Faster R-CNN | Reference standard caries status (expert annotation or clinical examination) | Specificity for caries detection Photographic images | Specificity 0.751 [0.646, 0.833] GRADE not reported 5 studies |
| Faster R-CNN | Reference standard caries status (expert annotation or clinical examination) | AUC for caries detection (HSROC) Photographic images | AUC 0.83 GRADE not reported 5 studies |
| Mask R-CNN | Reference standard caries status (expert annotation or clinical examination) | AUC for caries detection (HSROC) Photographic images | AUC 0.91 GRADE not reported 3 studies |
| Mask R-CNN | Reference standard caries status (expert annotation or clinical examination) | Sensitivity for caries detection Photographic images | Sensitivity 0.835 [0.673, 0.925] GRADE not reported 3 studies |
| Mask R-CNN | Reference standard caries status (expert annotation or clinical examination) | Specificity for caries detection Photographic images | Specificity 0.86 [0.705, 0.941] GRADE not reported 3 studies |
| Mask R-CNN (indirect comparison across different study sets) | Faster R-CNN | Sensitivity for caries detection All images | Pooled sensitivity Mask R-CNN vs Faster R-CNN 85.6% vs. 71.7%; indirect Z-test p = 0.0244; 6 Mask R-CNN and 11 Faster R-CNN studies GRADE not reported 17 studies |
| Mask R-CNN (indirect comparison across different study sets) | Faster R-CNN | Specificity for caries detection All images | Pooled specificity Mask R-CNN vs Faster R-CNN 94.2% vs. 81.4%; indirect Z-test p = 0.00089; 6 Mask R-CNN and 11 Faster R-CNN studies GRADE not reported 17 studies |
| Mask R-CNN (indirect comparison across different study sets) | Faster R-CNN | AUC for caries detection All images | Pooled auc Mask R-CNN vs Faster R-CNN 0.95 vs 0.84; DeLong test p = 0.0053; 6 Mask R-CNN and 11 Faster R-CNN studies GRADE not reported 17 studies |
| Mask R-CNN (indirect comparison across different study sets) | Faster R-CNN | Specificity for caries detection Radiographic images | Pooled specificity Mask R-CNN vs Faster R-CNN 96.5% vs. 85.0%; indirect Z-test p = 0.00105; 4 Mask R-CNN and 6 Faster R-CNN studies GRADE not reported 10 studies |
| Mask R-CNN (indirect comparison across different study sets) | Faster R-CNN | AUC for caries detection Radiographic images | Pooled auc Mask R-CNN vs Faster R-CNN 0.97 vs 0.86; DeLong test p = 0.0067; 4 Mask R-CNN and 6 Faster R-CNN studies GRADE not reported 10 studies |
| Mask R-CNN (indirect comparison across different study sets) | Faster R-CNN | Sensitivity for caries detection Radiographic images | Pooled sensitivity Mask R-CNN vs Faster R-CNN 86.3% vs. 67.2%; indirect Z-test p = 0.0497; 4 Mask R-CNN and 6 Faster R-CNN studies GRADE not reported 10 studies |
| Mask R-CNN (indirect comparison across different study sets) | Faster R-CNN | Specificity for caries detection Photographic images | Pooled specificity Mask R-CNN vs Faster R-CNN 86.0% vs. 75.1%; indirect Z-test p = 0.156; 3 Mask R-CNN and 5 Faster R-CNN studies GRADE not reported 8 studies |
| Mask R-CNN (indirect comparison across different study sets) | Faster R-CNN | Sensitivity for caries detection Photographic images | Pooled sensitivity Mask R-CNN vs Faster R-CNN 83.5% vs. 77.3%; indirect Z-test p = 0.435; 3 Mask R-CNN and 5 Faster R-CNN studies GRADE not reported 8 studies |
| Mask R-CNN (indirect comparison across different study sets) | Faster R-CNN | AUC for caries detection Photographic images | Pooled auc Mask R-CNN vs Faster R-CNN 0.91 vs 0.83; DeLong test p = 0.048; 3 Mask R-CNN and 5 Faster R-CNN studies GRADE not reported 8 studies |
| Mask R-CNN | Faster R-CNN (head-to-head on the same dataset) | Diagnostic accuracy for caries detection | no included study directly compared Faster R-CNN and Mask R-CNN on the same dataset under identical experimental conditions No studies found 0 studies |
Included primary studies 16
Open in Studies| Study | Year | Also included by | Outcomes it informs here | Trial data |
|---|---|---|---|---|
| Chen 2022 Detection of Proximal Caries Lesions on Bitewing Radiographs Using Deep Learning Method. PMID:36215971 | 2022 | 2 other reviews | not stated | not extracted yet |
| Moutselos 2019 Recognizing Occlusal Caries in Dental Intraoral Images Using Deep Learning. PMID:31946206 | 2019 | 1 other review | not stated | not extracted yet |
| Park 2022 Caries detection with tooth surface segmentation on intraoral photographic images using deep learning. PMID:36476359 | 2022 | 1 other review | not stated | not extracted yet |
| Cauas 2023 matched by author and year (probable) | NR | this review only | not stated | not extracted yet |
| Chaves 2024 Detection of caries around restorations on bitewings using deep learning. PMID:38342368 | 2024 | this review only | not stated | not extracted yet |
| Estai 2022 Evaluation of a deep learning system for automatic detection of proximal surface dental caries on bitewing radiographs. PMID:35534406 | 2022 | this review only | not stated | not extracted yet |
| Guo 2024 Rapid detection of non-normal teeth on dental X-ray images using improved Mask R-CNN with attention mechanism. PMID:38170416 | 2024 | this review only | not stated | not extracted yet |
| Juyal 2023 Dental Caries Detection Using Faster R-CNN and YOLO V3 DOI:10.1051/itmconf/20235302005 | 2023 | this review only | not stated | not extracted yet |
| Kunt 2023 Automatic caries detection in bitewing radiographs: part I-deep learning. PMID:37968358 | 2023 | this review only | not stated | not extracted yet |
| Liu 2020 A Smart Dental Health-IoT Platform Based on Intelligent Hardware, Deep Learning, and Mobile Terminal. PMID:31180873 | 2020 | this review only | not stated | not extracted yet |
| Mahaveerakannan 2024 matched by author and year (probable) | NR | this review only | not stated | not extracted yet |
| Rashid 2022 A hybrid mask RCNN-based tool to localize dental cavities from real-time mixed photographic images. PMID:35494840 | 2022 | this review only | not stated | not extracted yet |
| Thanh 2022 Deep Learning Application in Dental Caries Detection Using Intraoral Photos Taken by Smartphones DOI:10.3390/app12115504 | 2022 | this review only | not stated | not extracted yet |
| van Nistelrooij 2024 Deep Learning-Based Algorithm for Staging Secondary Caries in Bitewings. PMID:39471790 | 2025 | this review only | not stated | not extracted yet |
| Velusamy 2024 Faster Region‐based Convolutional Neural Networks with You Only Look Once multi‐stage caries lesion from oral panoramic X‐ray images DOI:10.1111/exsy.13326 | 2023 | this review only | not stated | not extracted yet |
| Zhu 2022 Faster-RCNN based intelligent detection and localization of dental caries DOI:10.1016/j.displa.2022.102201 | 2022 | this review only | not stated | not extracted yet |
Overlap with other reviews
Reviews that include the same primary studies are not independent evidence. Shared counts match studies by PMID, DOI or NCT, and otherwise by author and year (a probable match). Only the non-Cochrane dental reviews have study lists, so overlap with Cochrane reviews is not counted yet.
Review question
| Element | Planned (methods) | Found (results) |
|---|---|---|
| Population | Dental images (radiographic and/or photographic) for caries detection | Dental caries images, 41,384 tooth images (after augmentation; 90 to 12,750 per study); 17 studies |
| Intervention | Faster R-CNN and/or Mask R-CNN | Faster R-CNN (11 studies) and Mask R-CNN (6 studies) |
| Comparison | Reference standard or clearly defined alternative verification of caries status | Annotator or clinician reference standard (reporting varied) |
| Outcomes | Sensitivity, specificity, AUC (2x2 data TP/FP/TN/FN) | Pooled sensitivity, specificity and AUC by architecture and modality |
| Study designs | Cross-sectional or retrospective diagnostic accuracy studies; English only | Retrospective or cross-sectional diagnostic accuracy studies (17) |
| Timing | Not specified | Not specified |