Beyond visual inspection: can a multimodal machine learning model improve the preoperative differentiation of endometrial polyps from non-polypoid endometrial lesions?

The authors aimed to develop a multimodal machine learning model that integrates clinical data and ultrasound features to enhance the non-invasive preoperative differentiation between endometrial polyps and non-polypoid endometrial lesions. Their study found that while the gradient boosting decision tree model showed improved discriminative performance compared to traditional methods, the overall diagnostic accuracy was still insufficient for clinical decision-making, underscoring the need for histopathological confirmation. The findings suggest that further advancements, particularly in incorporating quantitative radiomic features, are necessary for better preoperative risk stratification.