Multimodal deep learning with anatomically constrained attention for screening MRI-detectable TMJ abnormalities from panoramic images - Summary - MDSpire
Advertisement
Multimodal deep learning with anatomically constrained attention for screening MRI-detectable TMJ abnormalities from panoramic images
To develop a comprehensive screening method for MRI-detectable temporomandibular joint (TMJ) pathologies, emphasizing the importance of early detection.
Key Findings:
The best-performing ensemble framework achieved an area under the curve of 0.86, indicating strong diagnostic performance.
Balanced classification of MRI-negative and -positive cases was achieved, which is crucial for accurate diagnosis.
Gradient-weighted Class Activation Mapping visualizations confirmed focus on condylar regions, supporting the framework's clinical relevance.
Ablation studies demonstrated the added value of clinical metadata and spatial attention, highlighting their importance in the diagnostic process.
Interpretation:
The prototype workflow can effectively triage TMJ patients for MRI referral, facilitating early detection and timely interventions for TMJ abnormalities.
Limitations:
Datasets are not publicly available due to patient privacy and ethical considerations, which may limit reproducibility and further research.
Conclusion:
The developed framework shows promise for improving early diagnosis of TMJ disorders, potentially leading to better patient outcomes.
