I assume that's just geometry and not parametrics to the veins etc? If that's the case then I'm afraid its all hard work. You can use the snapping tool to draw splines that stick to vertexes along the vein's path and then offset them (of super accuracy is needed for a centrally located spline in the middle of the vein).
If the vertex data is organized ( such that as you go down through the list of vertexes and you get clusters of each "ring" of vertexes that make up the vein tube). You could write some code to step through those clusters and find the average of the positions of all the points in each ring cluster and then place a new vertex at the middle average and then join all the new vertexes up.
There's an outside chance that if you scale each vein they may scale to a smaller diameter only. That would let you match to vertexes located centrally to the vein tube, but this is a very outside chance.