Skip to main content
Fig. 2 | Thrombosis Journal

Fig. 2

From: High-resolution magnetic resonance imaging-based radiomic features aid in selecting endovascular candidates among patients with cerebral venous sinus thrombosis

Fig. 2

Performance of the radiomic signature model. (a) A total of 13 RFs were selected based on least absolute shrinkage and selection operator (LASSO) regression. (b) The receiver operating characteristic (ROC) curve of the signature model in the training group. The optimal cutoff of the signature model was − 100.4. (c) The ROC curve of the signature model in the validation group. -100.4 was used as the optimal cutoff. (d) The distribution of clinic characteristics arranged according to increasing risk scores. (*, p < 0.05) (e–f) The distribution of risk scores stratified according to history of dyslipidemia and smoking. (g–i) The distribution of days from symptom onset, fibrin degradation product, and D-dimer levels stratified according to risk scores (*, p < 0.05). AUC: area under curve; BMI: body mass index; FDP: fibrin degradation product; ICP: intracranial hypertension; INR: international normalized ratio; ROC: receiver operating characteristic

Back to article page

Thrombosis Journal

ISSN: 1477-9560

Contact us