[MINC-users] Differences between mincstats and volume_stats

[Andrew Janke]

On Thu, 25 Sep 2003, Andre MATOS wrote:

> I was comparing the results from mincstats and the volume_stats and I
> found some different values, specially in Median. The mincstats was
> run into a Mac and the volume_stats into Linux. Below are both:
>
> mincstats sanca6_MTe2.mnc.gz -mask sanca6_2003_09_22_bayes1.mnc.gz
> -mask_range 4,4 -all
> ...
>
> Bus error

This error (I think) is a memory allocation problem that has been fixed in later
releases.  Which version of minc are you running?

> Is anyone tryed to compile the volume_stats (N3 package) into a Mac OS X
> and also know which program (mincstats or volume_stats) is right?

The answer is: Both are right.  However the two have a differing opinion of how
a histogram bin should be defined and also how many bins there should be in a
histogram.

  'man mincstats' will tell you how you can get mincstats to more closely
resemble volume_stats.  The base statistics should not differ (apart from via
rounding errors) but the histogram stats (of which median is one) are definitely
going to differ if you use the default options.

As for using mincstats instead of volume_stats in the mni_autoreg package, in
99% of cases there will be identical functionality.  N3 may well have
differences though as the authour of N3 is also the authour of volume_stats...


--
Andrew Janke   ( rotor@cmr.uq.edu.au || www.cmr.uq.edu.au/~rotor )
Australia->University of Queensland->Centre for Magnetic Resonance
W: +61 7 3365 4100  ||  H: +61 7 3800 4042  ||  M: +61 4 2138 8581