Interpretable CT machine learning model differentiates adrenal lipid-poor adenomas from pheochromocytomas

A decision tree model using multi-phase CT radiomic features achieved AUC 0.977 (95% CI 0.934–1.000) in lesions with washout >60% and 0.983 (95% CI 0.962–1.000) in those without, distinguishing adrenal lipid-poor adenomas from pheochromocytomas.

• The study included 229 patients with pathologically confirmed adrenal tumors, and models were trained on radiomic features from multi-phase contrast-enhanced CT.
• Cystic degeneration was the most influential discriminative feature, followed by enhancement potential and baseline attenuation.
• The decision tree model offered inherent interpretability with clear hierarchical decision rules.

Automated summary

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