[MINC-users] mincblur with FWHM =?utf-8?Q?<=3D_?=voxel dims

[Alex Zijdenbos]

Hrm those pesky licensing issues :-}

Not that I necessarily want to continue to harp on mincblur (where is Louis anyways?), but I just stumbled across something else with mincblur that has me a bit stumped. I had to dig this one fairly deep out of a registration script which seemed to be behaving oddly.

Take a look at this image: ​test_blur.png (http://cl.ly/2E362x310I231C1T1W3u)

These are two virtually identical images with the standard 1mm stx sampling (top two rows); the third row is the difference between them (simple subtraction); and the 4th and 5th rows are the result of mincblurring each of them with an 8mm FWHM kernel. For some reason, the blurred result of (only) one of the two produces a strong edge artifact.

Adding the -no_apodize option removed the edge effect; but that doesn't seem to explain why two almost-identical images would result in such different blurring behaviour.

-- A

On Tuesday, 26 June, 2012 at 3:21 AM, Andrew Janke wrote:

> On 26 June 2012 10:24, Vladimir S. Fonov <vladimir.fonov at gmail.com (mailto:vladimir.fonov at gmail.com)> wrote:
> > On the other hand mincblur is part of mni_autoreg with it's own
> > implementation of FFT
> >  
>  
>  
> My understanding is that louis based his fft upon the NR in C version.
> mincblur has also long been on my list of things to re-implement with
> support for arbitrary bluring in multiple dimensions (>4) but haven't
> got to it.
>  
> There is also mincfft that uses fftw but it is a separate package for
> this same licensing issue.
>  
> a
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca (mailto:MINC-users at bic.mni.mcgill.ca)
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
>  
>