Clustering different perfusion compartments in the brain is critical to the profound analysis of brain perfusion. This thesis presents a method based on a mixture of multivariate Gaussians (MoMG) and the expectation-maximization (EM) algorithm initialized by the results of hierarchical clustering (HC) on the whitened data to automatically dissect various perfusion compartments from dynamic susceptibility contrast (DSC) MR images so that each compartment comprises pixels of similar signal-time curves. This EM-HC-based method provides an objective means to 1) delineate an area to serve as the in-plane arterial input function (AIF) of feeding artery for the adjacent tissues to better quantify the relative cerebral blood volume (rCBV), relative cerebral blood flow (rCBF), and mean transit time (MTT); 2) demarcate regions with abnormal perfusion derangement and facilitate diagnosis; and 3) provide parametric maps with supplementary information, such as temporal scenarios and recirculation of contrast agent. Monte Carlo simulations have been designed and conducted to assess the performance of the proposed method and investigate the acceptable temporal and spatial resolutions under various noise levels. Results from normal subjects show that perfusion cascade manifests, in order of appearance, the artery, gray matter, white matter, vein and sinus, and choroid plexus mixed with cerebrospinal fluid (CSF). The averaged rCBV, rCBF, and MTT ratios between gray matter and white matter are in good agreement with those in the literature. Results on clinical cases have shown distinct spatiotemporal characteristics between perfusion patterns, differentiating pathological areas from nonpathological ones.