CT based 3D radiomic and clinical airway examination model for evaluating mask ventilation in oral and maxillofacial surgery patients.

Summary

In 716 oral/maxillofacial surgery patients, a mixed model integrating five 3D CT radiomic signatures with six clinical measures predicted difficult mask ventilation with an AUC of 0.851, outperforming clinical-only and radiomics-only models.

Key Findings

• Prospective cohort of 716 patients with standardized grading of mask ventilation difficulty.
• Mixed model combining five 3D radiomic signatures and six clinical features achieved AUC 0.851 in validation, outperforming single-source models.
• Difficult mask ventilation defined pragmatically (need for assistance or adjuvants), aligning with clinical relevance.

Clinical Implications

In patients already undergoing head/neck CT, integrating radiomics with bedside assessments can flag difficult mask ventilation, prompting early adjunct preparation, backup plans, and senior support.

Limitations

• Single-center setting; lacks external validation and calibration across scanners/protocols
• Relies on preoperative CT availability, limiting generalizability beyond maxillofacial cohorts

Future Directions

External, multi-center validation; assess impact on airway-related adverse events and workflow; explore integration with ultrasound or video data and automated segmentation.