A local maximum intensity projection (MIP) approach to the extraction of a 3D vascular network, acquired by the KnifeEdge Scanning Microscope (KESM), is presented. We build a local volume for local MIP processing at each tracing step in order to reduce the dimension of input data from 3D to 2D, which leads to a 65.22% reduction of computation time compared to 3D tracing method. The proposed method makes use of existing 2D tracing methods, extending them into a 3D tracing method. Our experimental results show that our approach can rapidly and accurately extract the medial axis of vascular data acquired by the KESM.