Non-Cochrane · DTA · PM41485242
Classification of Salivary Gland Tumors on Ultrasound Using Artificial Intelligence: A Systematic Review and Meta-Analysis.
Chau Isabelle J, Kim Cory Hyun-Su, Nguyen Shaun A, et al. · Otolaryngology--head and neck surgery : official journal of American Academy of Otolaryngology-Head and Neck Surgery · 2026 · PubMed · doi:10.1002/ohn.70102
Findings
6 findings
| Population | ||||
|---|---|---|---|---|
| Patients with salivary gland tumors evaluated by ultrasound | Artificial intelligence models applied to ultrasound images | Histopathology reference standard (final histopathology or core needle biopsy) | Pooled specificity for classification of benign vs malignant Test cohorts | Pooled specificity 0.86 [0.8, 0.92] GRADE not reported 9 studies · 772 participants |
| Patients with salivary gland tumors evaluated by ultrasound | Artificial intelligence models applied to ultrasound images | Histopathology reference standard (final histopathology or core needle biopsy) | Pooled accuracy for classification of benign vs malignant Test cohorts | Pooled accuracy 0.85 [0.81, 0.9] GRADE not reported 9 studies · 772 participants |
| Patients with salivary gland tumors evaluated by ultrasound | Artificial intelligence models applied to ultrasound images | Histopathology reference standard (final histopathology or core needle biopsy) | Pooled sensitivity for classification of benign vs malignant Test cohorts | Pooled sensitivity 0.91 [0.86, 0.95] GRADE not reported 9 studies · 772 participants |
| Patients with parotid pleomorphic adenoma or Warthin tumor evaluated by ultrasound | Artificial intelligence models applied to ultrasound images | Histopathology reference standard (postoperative histopathology) | Pooled specificity for classification of pleomorphic adenoma vs Warthin tumor Test cohorts | Pooled specificity 0.88 [0.81, 0.95] GRADE not reported 3 studies · 135 participants |
| Patients with parotid pleomorphic adenoma or Warthin tumor evaluated by ultrasound | Artificial intelligence models applied to ultrasound images | Histopathology reference standard (postoperative histopathology) | Pooled accuracy for classification of pleomorphic adenoma vs Warthin tumor Test cohorts | Pooled accuracy 0.84 [0.79, 0.9] GRADE not reported 3 studies · 135 participants |
| Patients with parotid pleomorphic adenoma or Warthin tumor evaluated by ultrasound | Artificial intelligence models applied to ultrasound images | Histopathology reference standard (postoperative histopathology) | Pooled sensitivity for classification of pleomorphic adenoma vs Warthin tumor Test cohorts | Pooled sensitivity 0.81 [0.74, 0.89] GRADE not reported 3 studies · 135 participants |
Included primary studies 13
Open in Studies| Study | Year | Also included by | Outcomes it informs here | Trial data |
|---|---|---|---|---|
| Cheng 2023 Diagnosis of Salivary Gland Tumors Using Transfer Learning with Fine-Tuning and Gradual Unfreezing. PMID:37958229 | 2023 | this review only | not stated | not extracted yet |
| He 2024 An ultrasound-based ensemble machine learning model for the preoperative classification of pleomorphic adenoma and Warthin tumor in the… PMID:38570381 | 2024 | this review only | not stated | not extracted yet |
| Jiang 2024 Deep learning-assisted diagnosis of benign and malignant parotid tumors based on ultrasound: a retrospective study. PMID:38654281 | 2024 | this review only | not stated | not extracted yet |
| Li 2022 A nomogram based on clinical information, conventional ultrasound and radiomics improves prediction of malignant parotid gland lesions. PMID:34929334 | 2022 | this review only | not stated | not extracted yet |
| Liu 2024 Deep learning based ultrasound analysis facilitates precise distinction between parotid pleomorphic adenoma and Warthin tumor. PMID:38476360 | 2024 | this review only | not stated | not extracted yet |
| Mao 2024 Multi-feature Fusion Network on Gray Scale Ultrasonography: Effective Differentiation of Adenolymphoma and Pleomorphic Adenoma. PMID:38871552 | 2024 | this review only | not stated | not extracted yet |
| Shan 2025 Machine Learning Differentiates Between Benign and Malignant Parotid Tumors With Contrast-Enhanced Ultrasound Features. PMID:39557074 | 2025 | this review only | not stated | not extracted yet |
| Tu 2023 Neural network combining with clinical ultrasonography: A new approach for classification of salivary gland tumors. PMID:37222027 | 2023 | this review only | not stated | not extracted yet |
| Wang 2022 The Diagnostic Value of Ultrasound-Based Deep Learning in Differentiating Parotid Gland Tumors DOI:10.1155/2022/8192999 | 2022 | this review only | not stated | not extracted yet |
| Wang 2022 The Diagnostic Value of Ultrasound-Based Deep Learning in Differentiating Parotid Gland Tumors. PMID:35602298 | 2022 | this review only | not stated | not extracted yet |
| Wang 2024 Differentiation of benign and malignant parotid gland tumors based on the fusion of radiomics and deep learning features on ultrasound… PMID:38803533 | 2024 | this review only | not stated | not extracted yet |
| Wei 2024 Deep learning-assisted diagnosis of benign and malignant parotid gland tumors based on automatic segmentation of ultrasound images: a… PMID:39184051 | 2024 | this review only | not stated | not extracted yet |
| Zhang 2023 A deep learning model for the differential diagnosis of benign and malignant salivary gland tumors based on ultrasound imaging and… PMID:37179911 | 2023 | this review only | not stated | not extracted yet |
Review question
| Element | Planned (methods) | Found (results) |
|---|---|---|
| Population | Patients of all ages with salivary gland lesions evaluated by ultrasound | Patients with salivary gland tumors (N=4721; 12 studies; mostly parotid; China n=10, Taiwan n=2) |
| Intervention | Artificial intelligence (machine learning, deep learning, radiomics) combined with ultrasound to classify salivary gland tissue | AI models with conventional ultrasound (11 studies) or contrast-enhanced ultrasound (1 study): deep learning (8), machine learning (2), combined machine and deep learning (1), radiomics (1) |
| Comparison | Traditional diagnostic methods; histopathology as reference standard | Final histopathology (postoperative or core needle biopsy) as reference standard |
| Outcomes | Accuracy, sensitivity, specificity (or confusion matrix) | Pooled sensitivity, specificity, accuracy in test cohorts |
| Study designs | Cohort studies, randomized controlled trials, case series | Retrospective cohort studies, Oxford level of evidence 2 (n=12) |
| Timing | Not specified | Not specified |