Combined DWI habitat and Transformer model predicts progesterone receptor status in meningiomas

A combined model integrating DWI-based habitat analysis and Transformer deep learning predicted progesterone receptor expression in meningiomas with AUC 0.910 (training) and 0.902 (external test), outperforming the clinical model.

• The model combined intratumoral heterogeneity features from DWI habitat analysis with a Transformer-based deep learning model on the same images.
• It was validated in an independent external test set from a second center, showing robust performance (AUC 0.902).
• The combined model significantly outperformed a model using only clinical and conventional imaging features.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.