Identifying sinonasal inverted papilloma by machine learning: a systematic review and meta-analysis

Area of Science:

• Otolaryngology
• Medical Imaging
• Artificial Intelligence

Background:

• Sinonasal inverted papilloma (IP) is a benign sinonasal tumor with malignant potential.
• Accurate diagnosis of IP is crucial for appropriate management.
• Machine learning (ML) has not been previously evaluated for IP diagnosis.

Purpose of the Study:

• To systematically review and perform a meta-analysis on the diagnostic performance of ML for sinonasal inverted papilloma (IP).
• To assess the accuracy of ML models in differentiating IP from malignant tumors.

Main Methods:

• Systematic literature search of PubMed, Cochrane, Embase, and Web of Science databases.
• Quality assessment of included studies using the QUADAS-2 tool.
• Meta-analysis using a bivariate mixed-effect model.

Main Results:

• 17 studies with 3321 participants were included.
• ML models based on radiomics achieved a sensitivity of 0.84 and specificity of 0.82.
• ML models incorporating radiomics and clinical features showed improved sensitivity (0.85) and specificity (0.87).

Conclusions:

• Machine learning demonstrates favorable diagnostic performance for the differential diagnosis of IP.
• Further prospective studies are recommended to validate ML tools for broader clinical application.