HR-VWI radiomics model predicts recanalization in cervical artery dissection

In 110 patients with cervical artery dissection, a logistic regression model combining HR-VWI radiomics and clinical factors achieved an AUC of 0.757 (95% CI 0.531-0.983) for predicting vascular recanalization, though limited sample size and few events warrant caution.

• The best-performing model integrated 6 radiomic features from high-resolution vessel wall imaging (HR-VWI) with 5 clinical factors (hyperuricemia, hyperlipidemia, smoking, headache, stenosis degree).
• The logistic regression model showed moderate discriminative ability with an AUC of 0.757 and acceptable calibration (Brier score 0.129) in an independent test set.
• Given the small sample and few recanalization events, the authors describe this as a proof-of-concept requiring external validation before clinical use.

Automated summary

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