Lee, the code is okay, when Lambda1 and Lambda2 are passed from eig2image their values are actually switched (check line 108 in FrangiFilter2D).
My question would be, has anyone tried testing the 3D algorithm on a series of MRI scans?
I take 56 scans, load them into one array but can't seem to get the vessels out. Could this be because the resolution between slices is greater then intra-slice resolution?
Any idea is greatly appreciated.