Redundancy-aware radiomics on multiparametric MRI distinguishes melanoma from NSCLC brain metastases

A radiomics model on multiparametric MRI achieves an AUC of 0.75 for distinguishing melanoma from non-small cell lung cancer (NSCLC) brain metastases, with FLAIR and T2 sequences contributing most to classification, according to a new study.

• The best-performing model used a random forest with 200 features (RF-200) and achieved a test AUC of 0.75; SHAP analysis revealed that fluid-sensitive sequences (FLAIR, T2) dominated modality contributions, capturing heterogeneity beyond contrast enhancement.
• No external validation or prospective reader comparison was performed, limiting immediate clinical applicability.

Automated summary

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