On Mon, Sep 10, 2012 at 11:21:02AM -0700, Jonas Kaplan wrote: > Seems to me that radius of 1 gives 7 voxels, e.g.: > > In [117]: radii = range(5) > > In [118]: for radius in radii: > .....: sp = Sphere(radius=radius) > .....: size = len(sp((1,1,1))) > .....: print "Radius %d size: %d" % (radius,size) > .....: > Radius 0 size: 1 > Radius 1 size: 7 > Radius 2 size: 33 > Radius 3 size: 123 > Radius 4 size: 257
Thanks Jonas! In PyMVPA >= 0.6 voxel neighbors are elements that share an edge (not just a corner), hence 7 elements for radius=1 (2 elements per axis and the center). radius=0 will select just the center. When selecting neighbors in a real (masked) dataset, the actual number of elements can be less than the theoretical maximum, whenever not all the potential neighbor coordinates are actually part of the dataset/mask. Hope that explains, Michael -- Michael Hanke http://mih.voxindeserto.de _______________________________________________ Pkg-ExpPsy-PyMVPA mailing list [email protected] http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa
