Because of its human and economic cost, there is a need to reduce the disability resulting from a stroke. Though emergency reperfusion treatment can reduce disability, complications resulting from it, such as intracerebral hemorrhage, can be lethal. Many patients suffering from ischemic stroke develop varying degrees of pial-collateral arterial supply (PAS), which can affect patient response to reperfusion therapy and the risk of developing intracerebral hemorrhage. Observation of good PAS predicts a more favorable outcome (reduced disability) when performing reperfusion treatment. ,Current methods for assessing pial collaterals use either (a) physically-installed pial windows, or (b) manual scoring of the extent of PAS on X-ray digital subtraction angiography (DSA) image series. Though pial windows provide microscopic visualization of pial collaterals, during a stroke, this method is clinically infeasible. Manual scoring off of X-ray DSA series is far more preferable because of X-ray DSA's availability during intervention, its resolution and scan time, and because manual scoring techniques can be reproducible and quantitative. However, these techniques' ultimate performances are coarse and dependent on viewer experience. Therefore, the objective of this dissertation is to investigate and develop a computational method to quantitatively assess PAS---imaged using X-ray DSA---in the setting of acute ischemic stroke. It is hypothesized that computerized and quantitative angiographic image analysis of pial arterial supply can be used to identify patients' suitability for reperfusion treatment. ,Digitally-subtracted angiograms were retrospectively collected under an institutional review board-approved, protocol compliant with the Health Insurance Portability and Accountability Act of 1996 (HIPAA). Occlusion sites included the M1 segment of the Middle Cerebral Artery for 15 patients, the proximal M2 segment for 1 patient, and the Internal Carotid Artery for 8 patients. Eleven of the patients were imaged at 6 Frames/second, while the remainder were imaged under an X-ray dose-sparing protocol.,The research in this dissertation covers 3 major topics. First, a Fuzzy C-Means (FCM) based approach for automatically segmenting PAS-affected vessels from capillary blush and background in X-ray DSA series during acute ischemic stroke was developed. With an area under the ROC curve of up to 0.89 across multiple frame-rates for the task of segmenting vessels from non-vessels, this method was shown to have robust performance and could identify vessels almost as well as an expert observer. Next, a quantitative method for extracting 10 features from kinetic contrast curves in X-ray DSA series and validating these features was developed. These features' abilities to distinguish between patients with favorable PAS from those with poorer PAS was evaluated. For the task of identifying patients with particularly poor PAS, many of these features had areas under the Receiver Operating Characteristic (ROC) curves of approximately $0.99$, indicating a substantial capability for contra-indicating reperfusion treatment. Finally, a Fuzzy C-Means based approach for automatically segmenting arteries from non-arteries was developed and evaluated. Kinetic features were subsequently extracted from curves generated from segmented arteries and segmented parenchymal blush due to capillaries, and their performances in the task of distinguishing between patients with favorable PAS from those with poorer PAS was evaluated. The results suggested that FCM could segment arteries from non-arteries. Features extracted from pixel segmented arterial and capillary curves were comparable to features extracted without any segmentation; however, a mild improvement in performance for 2 features suggest that extracting features from arterial or capillary filling could provide real-time quantitative markers for a patient's condition during intervention.,The results support the hypothesis that during acute ischemic stroke, computerized and quantitative angiographic analysis of PAS can identify patients' suitability for reperfusion treatment. Therefore, this method can potentially serve as a fast ``second-check" to an interventionalist's treatment decisions, leading to better outcomes and reduced disability. Limitations that must be overcome prior to clinical adoption include the small size of the database (24 cases total), the vascular overlap caused by projecting a 3D spatial volume to a 2D spatial image, and the uncontrolled environment of clinical exams. These limitations can be addressed by including more cases, adapting the methods discussed in this dissertation to 4D (3D plus time) DSA series of acute ischemic stroke, and complementing the clinical studies presented in this thesis with simulation and preclinical canine studies, respectively.,This dissertation serves as the first step in achieving a fully computerized and quantitative means for personalizing patient management in intervention for acute ischemic stroke. Moreover, the techniques presented in this thesis may find application in quantifying imaging of other neurovascular disease.