High-resolution black-blood MRA images of intracranial vascular anatomy can be obtained using 3D fast spin-echo techniques. Although these images demonstrate excellent contrast between vessels and surrounding soft tissues, the dark signal from air and bone can obscure the desired vascular information when a minimum intensity projection image is created. In this paper, we describe an image processing technique based upon a median filter that is effective for detecting narrow vessel-like structures. Minimum intensity projection images of the filtered MRA volume can be obtained in any orientation without prior segmentation of the skull or surrounding air spaces. The filter is very effective for detecting and visualizing small vessels, but is much less effective for detecting vessels and vascular pathology larger than the filter detection width. The filtering technique is demonstrated on black-blood MRA images from a volunteer study.