Abstract: Disclosed herein are methods and systems for the identification and characterization of perivascular spaces in the cerebral vasculature using magnetic resonance imaging (MRI) data. The disclosed methods allow for automated and unbiased quantification of enlarged perivascular space (ePVS) in a subject, and thus can provide a substantial improvement over manual grading methods used in the art. An example method includes receiving an MRI dataset including voxels having intensity values, identifying a set of candidate voxels within the MRI dataset based on the intensity values; grouping the set of candidate voxels into a set of first clusters; filtering the set of first clusters to generate a set of second clusters; and filtering the set of second clusters based on a morphologic constraint, thereby identifying an enlarged perivascular space in the MRI dataset.

Abstract: Disclosed herein are methods and systems for the identification and characterization of perivascular spaces in the cerebral vasculature using magnetic resonance imaging (MRI) data. The disclosed methods allow for automated and unbiased quantification of enlarged perivascular space (ePVS) in a subject, and thus can provide a substantial improvement over manual grading methods used in the art. An example method includes receiving an MRI dataset including voxels having intensity values, identifying a set of candidate voxels within the MRI dataset based on the intensity values; grouping the set of candidate voxels into a set of first clusters; filtering the set of first clusters to generate a set of second clusters; and filtering the set of second clusters based on a morphologic constraint, thereby identifying an enlarged perivascular space in the MRI dataset.