Dear Khalil and Thomas,

The two volumes are actually not identical, the first one is continuous
while the second is a discretised volumetric approximation of a sphere.

For simple continuous shapes such as a sphere, there is a known formulae
to compute the Resel counts, see section 3.2 and table 1 p.18 of this
paper from Keith Worsley:
  http://www.math.mcgill.ca/keith/unified/unified.pdf
otherwise, for voxel data, the Resel counts is computed using equations
from section 3.3 of the same paper.

Have a look at spm_resels.m for the implementation.

This is also described in section 3.5 of this chapter from the SPM book:
  http://www.fil.ion.ucl.ac.uk/spm/doc/books/hbf2/pdfs/Ch14.pdf

Best regards,
Guillaume.


On 25/04/17 17:12, Khalil I Thompson wrote:
> Hey Thomas!
>
>
> I just ran across this issue with my data as well a couple of weeks ago.
> I believe the problem was that the ROI sphere will always cover the
> volume you specified when originally designing the mask so centering
> during VOI time-series extraction doesn't make any difference, the
> volume location is always consistent. The SVC correction, however, is
> completely dependent on the centered location so it won't necessarily
> overlap with the ROI sphere you originally designed. Does that make sense?
>
>
> Khalil Thompson
>
> Graduate Student
>
> Cognitive & Affective Neuroscience
>
> Georgia State University
>
> [log in to unmask]
>
> ------------------------------------------------------------------------
> *From:* SPM (Statistical Parametric Mapping) <[log in to unmask]> on
> behalf of Arnold, Thomas <[log in to unmask]>
> *Sent:* Tuesday, April 25, 2017 10:01:24 AM
> *To:* [log in to unmask]
> *Subject:* [SPM] SVC Resel discrepancy
>
>
> Greetings,
>
>
> I've performed small-volume correction using the a 6mm sphere as well as
> using an ROI image of a 6mm sphere. I noted that the P-values differed
> between the methods. Both spheres are centered at the same location and
> have the same FWHM values, voxel number, and voxel size. Despite this
> the resels are different (ROI is half that of sphere). Why would this be
> the case if the resels are a function of volume and FWHM, which are the
> same in both cases?
>
>
> Thank you for any help,
> Thomas Campbell Arnold
>

--
Guillaume Flandin, PhD
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG