Towards Automatic Collateral Circulation Score Evaluation in Ischemic Stroke Using Image Decompositions and Support Vector Machines.

Stroke is the second leading cause of disability worldwide. Thrombectomy has been shown to offer fast and efficient reperfusion with high recanalization rates and thus improved patient outcomes. One of the most important indicators to identify patients amenable to thrombectomy is evidence of good collateral circulation. Currently, methods for evaluating collateral circulation are generally limited to visual inspection with potentially high inter- and intra-rater variability. In this work, we present an automatic technique to evaluate collateral circulation. This is achieved via low-rank decomposition of the target subject’s 4D CT angiography, and using principal component analysis (PCA) and support vector machines (SVMs) to automatically generate a collateral circulation score. With the proposed automatic score evaluation technique, we have achieved an overall scoring accuracy of 82.2% to identify patients with poor, intermediate, and good/normal collateral circulation.