Generating whole-brain vein segmentations can be very time-consuming. In this abstract, a method is proposed that can segment brain veins from a single-echo or multi-echo gradient echo scan. The segmentation algorithm combines classical vessel enhancement filtering and local thresholding methods with a shearlet-based multi-scale approach. Compared with a ground truth, the algorithm performs better for multi-echo data when R2* information is included in the segmentation. Moreover, the combination of venous segmentation with masks for deep and superficial venous territories yields higher susceptibility values for the superficial venous vasculature which is in accordance with a higher oxygen consumption of the cortex.