CT Radiomics–Based ML Nomogram Predicts GPC3 in HCC Before Surgery

Combining CT radiomics with alpha-fetoprotein and total bilirubin yielded an AUC of 0.948 for predicting glypican-3 expression in hepatocellular carcinoma, up from 0.862 with radiomics alone.

• Retrospective two-center study in 103 patients used a random forest classifier with 10 radiomic features; training (n=72) and testing (n=31) sets.
• The combined nomogram showed positive net reclassification and integrated discrimination improvement, with decision curve analysis demonstrating greater net clinical benefit.

Automated summary

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