Reliable assessment of collateral blood supply is important in acute ischemic stroke. We propose a quantitative method for evaluation of collateral status on CT angiography (CTA). We collected CTA images of 70 patients from MR CLEAN with an occlusion in the M1 branch. Our proposed quantitative collateral score (qCS) consisted of atlas-based territory-at-risk identification and vessel segmentation using a Hessian-based filter. Ground truth was obtained by manual collateral scoring (mCS). Accuracy was evaluated by analysis of Spearman ρ and one-way ANOVA. Correlation of mCS and qCS with tissue death and functional outcome was assessed. Receiver operating characteristics curves of mCS and qCS were analyzed to distinguish favorable from unfavorable outcome. qCS strongly correlated with mCS and showed reliable correlations with tissue death and functional outcome. qCS showed higher discriminative power between favorable and unfavorable compared to mCS, indicating potential clinical value.

, , , ,
doi.org/10.1007/978-3-319-67564-0_18, hdl.handle.net/1765/102070
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Erasmus MC: University Medical Center Rotterdam

Boers, A., Sales Barros, R. (Renan), Jansen, I.G.H. (Ivo G. H.), Slump, C., Dippel, D., van der Lugt, A., … on behalf of the MR CLEAN Investigators, C.I. (CLEAN Investigators). (2017). Quantitative collateral grading on CT angiography in patients with acute ischemic stroke. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). doi:10.1007/978-3-319-67564-0_18