DriverNet MRI Model Predicts EGFR/ALK Mutations in NSCLC Brain Metastases Across Cohorts

DriverNet MRI model predicted EGFR/ALK mutation status in NSCLC brain metastases with AUCs of 0.952–0.962 in external test cohorts, outperforming clinical-only models. Prospective validation is needed.

• Multicenter study with 374 patients from three centers, including internal and external validation.
• DriverNet used T1CE and T2-FLAIR MRI with a Transformer-based architecture, outperforming clinical-only and imaging-only models.
• Model-derived labels were associated with overall survival in exploratory analysis; prospective validation in larger, balanced cohorts is needed.

Automated summary

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