Non-Cochrane · DTA · PM41655403
Artificial Intelligence-assisted Diagnosis of Carotid Artery Calcifications on Panoramic Radiographs: A Meta-analysis.
Ye Fangfei, Zhou Qun, Zhang Siying · International dental journal · 2026 · PubMed · doi:10.1016/j.identj.2026.109421
Findings
36 findings
| Population | ||||
|---|---|---|---|---|
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Negative likelihood ratio for carotid artery calcification diagnosis (per-side) Sensitivity analysis: best-performing model per study | NLR 0.11 [0.05, 0.23] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Diagnostic odds ratio for carotid artery calcification diagnosis (per-side) Number of reference specialists: 1 | DOR 392.6 [234.15, 658.28] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Sensitivity for carotid artery calcification diagnosis (per-side) AI task type: detection | Sensitivity 0.87 [0.74, 0.94] GRADE not reported 4 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Specificity for carotid artery calcification diagnosis (per-side) Number of reference specialists: 1 | Specificity 0.96 [0.95, 0.97] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Specificity for carotid artery calcification diagnosis (per-side) Dataset size: >=500 images | Specificity 0.93 [0.90, 0.96] GRADE not reported 7 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Sensitivity for carotid artery calcification diagnosis (per-side) Dataset size: >=500 images | Sensitivity 0.85 [0.78, 0.91] GRADE not reported 7 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Diagnostic odds ratio for carotid artery calcification diagnosis (per-side) Dataset size: >=500 images | DOR 83.72 [39.21, 178.72] GRADE not reported 7 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Diagnostic odds ratio for carotid artery calcification diagnosis (per-side) Dataset size: <500 images | DOR 210.12 [40.31, 1095.30] GRADE not reported 7 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Diagnostic odds ratio for carotid artery calcification diagnosis (per-side) AI task type: detection | DOR 38.46 [7.16, 206.58] GRADE not reported 4 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Sensitivity for carotid artery calcification diagnosis (per-side) | Sensitivity 0.88 [0.84, 0.92] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Diagnostic odds ratio for carotid artery calcification diagnosis (per-side) Sensitivity analysis: best-performing model per study | DOR 109 [22, 547] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Sensitivity for carotid artery calcification diagnosis (per-side) Number of reference specialists: 1 | Sensitivity 0.94 [0.91, 0.96] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Specificity for carotid artery calcification diagnosis (per-side) Number of reference specialists: >=2 | Specificity 0.91 [0.84, 0.95] GRADE not reported 8 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Positive likelihood ratio for carotid artery calcification diagnosis (per-side) Sensitivity analysis: best-performing model per study | PLR 11.7 [4.6, 29.7] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Sensitivity for carotid artery calcification diagnosis (per-side) Sensitivity analysis: best-performing model per study | Sensitivity 0.90 [0.81, 0.95] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Diagnostic odds ratio for carotid artery calcification diagnosis (per-side) Number of reference specialists: >=2 | DOR 42.64 [21.71, 83.72] GRADE not reported 8 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Positive likelihood ratio for carotid artery calcification diagnosis (per-side) | PLR 15.8 [9.0, 27.6] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Specificity for carotid artery calcification diagnosis (per-side) AI task type: classification | Specificity 0.97 [0.95, 0.98] GRADE not reported 9 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Negative likelihood ratio for carotid artery calcification diagnosis (per-side) | NLR 0.12 [0.08, 0.18] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Sensitivity for carotid artery calcification diagnosis (per-side) Dataset size: <500 images | Sensitivity 0.91 [0.83, 0.95] GRADE not reported 7 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Area under the SROC curve for carotid artery calcification diagnosis (per-side) | AUC 0.96 [0.94, 0.98] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Sensitivity for carotid artery calcification diagnosis (per-side) Number of reference specialists: >=2 | Sensitivity 0.82 [0.76, 0.86] GRADE not reported 8 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Diagnostic odds ratio for carotid artery calcification diagnosis (per-side) AI task type: classification | DOR 232.2 [88.59, 608.62] GRADE not reported 9 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Specificity for carotid artery calcification diagnosis (per-side) Dataset size: <500 images | Specificity 0.96 [0.88, 0.99] GRADE not reported 7 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Specificity for carotid artery calcification diagnosis (per-side) AI task type: detection | Specificity 0.85 [0.70, 0.94] GRADE not reported 4 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Diagnostic odds ratio for carotid artery calcification diagnosis (per-side) | DOR 129 [55, 305] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Sensitivity for carotid artery calcification diagnosis (per-side) AI task type: classification | Sensitivity 0.89 [0.82, 0.93] GRADE not reported 9 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Specificity for carotid artery calcification diagnosis (per-side) Sensitivity analysis: best-performing model per study | Specificity 0.92 [0.82, 0.97] GRADE not reported 6 studies |
| Panoramic radiographs (left and right sides evaluated separately) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Specificity for carotid artery calcification diagnosis (per-side) | Specificity 0.94 [0.91, 0.97] GRADE not reported 6 studies |
| Panoramic radiographs (person-based evaluation) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Diagnostic odds ratio for carotid artery calcification diagnosis (per-person) | DOR 37 [10, 144] GRADE not reported 2 studies |
| Panoramic radiographs (person-based evaluation) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Negative likelihood ratio for carotid artery calcification diagnosis (per-person) | NLR 0.12 [0.04, 0.36] GRADE not reported 2 studies |
| Panoramic radiographs (person-based evaluation) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Sensitivity for carotid artery calcification diagnosis (per-person) | Sensitivity 0.90 [0.74, 0.97] GRADE not reported 2 studies |
| Panoramic radiographs (person-based evaluation) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Area under the SROC curve for carotid artery calcification diagnosis (per-person) | AUC 0.86 [0.82, 0.88] GRADE not reported 2 studies |
| Panoramic radiographs (person-based evaluation) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Specificity for carotid artery calcification diagnosis (per-person) | Specificity 0.81 [0.73, 0.87] GRADE not reported 2 studies |
| Panoramic radiographs (person-based evaluation) | AI models (machine learning or deep learning) for carotid artery calcification detection on panoramic radiographs | Reference standard for carotid artery calcifications established by human clinicians | Positive likelihood ratio for carotid artery calcification diagnosis (per-person) | PLR 4.6 [3.1, 7.0] GRADE not reported 2 studies |
| Panoramic radiographs, per-side evaluation, best performing model per study | AI models (machine or deep learning) for detecting or classifying carotid artery calcifications | Reference standard for carotid artery calcifications established by clinical specialists | Area under the SROC curve Panoramic radiography; retrospective image datasets · Best model per study | AUC 0.96 [0.94, 0.98] GRADE not reported 6 studies |
Included primary studies 9
Open in Studies| Study | Year | Also included by | Outcomes it informs here | Trial data |
|---|---|---|---|---|
| Amitay 2023 Deep convolution neural network for screening carotid calcification in dental panoramic radiographs. PMID:37043433 | 2023 | this review only | not stated | not extracted yet |
| Kats 2019 Atherosclerotic carotid plaque on panoramic radiographs: neural network detection. PMID:31134222 | 2019 | this review only | not stated | not extracted yet |
| Kuwada 2025 Area detection improves the person-based performance of a deep learning system for classifying the presence of carotid artery… PMID:40694246 | 2026 | this review only | not stated | not extracted yet |
| Lee 2022 Detecting 17 fine-grained dental anomalies from panoramic dental radiography using artificial intelligence. PMID:35338198 | 2022 | this review only | not stated | not extracted yet |
| Sawagashira 2013 An Automatic Detection Method for Carotid Artery Calcifications Using Top-Hat Filter on Dental Panoramic Radiographs DOI:10.1587/transinf.e96.d.1878 | 2013 | this review only | not stated | not extracted yet |
| Sawagashira 2013 An automatic detection method for carotid artery calcifications using top-hat filter on dental panoramic radiographs. PMID:22255757 | 2011 | this review only | not stated | not extracted yet |
| Song 2022 Comparison of detection performance of soft tissue calcifications using artificial intelligence in panoramic radiography. PMID:36352043 | 2022 | this review only | not stated | not extracted yet |
| Vinayahalingam 2024 Detection of carotid plaques on panoramic radiographs using deep learning. PMID:39461583 | 2024 | this review only | not stated | not extracted yet |
| Yoo 2024 CACSNet for automatic robust classification and segmentation of carotid artery calcification on panoramic radiographs using a cascaded… PMID:38886356 | 2024 | this review only | not stated | not extracted yet |
Review question
| Element | Planned (methods) | Found (results) |
|---|---|---|
| Population | Panoramic radiographs | Panoramic radiographs from 8 retrospective studies (datasets of 60 to 1400 images); per-side (n=6 studies) and per-person (n=2 studies) analyses |
| Intervention | Any AI model (machine or deep learning) for automatic detection, classification or segmentation of carotid artery calcifications on panoramic radiographs | 13 AI models (deep learning mostly; 1 SVM machine learning study); classification and detection tasks |
| Comparison | Reference standard for carotid artery calcifications established by human clinicians | Reference standard by 1 to 4 clinical specialists (radiologists, oral and maxillofacial specialists, dentists) |
| Outcomes | Diagnostic performance: sensitivity, specificity, AUC, PPV, NPV, DOR and related measures | Sensitivity, specificity, PLR, NLR, DOR, AUC (7 studies in meta-analysis) |
| Study designs | Not specified | Retrospective diagnostic accuracy studies (n=8; 7 in meta-analysis) |
| Timing | Not specified | Not specified |