[MINC-users] Co-registration metric (minctracc)

[Andrew Janke]

Not as of yet. Looking at the code it would seem that the floor value
only applies to the first input volume.

   https://github.com/BIC-MNI/minc/blob/master/progs/minccmp/minccmp.c#L333

It would not be a hard change to make, the code for dealing with
ranges of input values is already in mincstats.

For now I'd suggest a first pass through with minccalc setting values
in the first to 0 if the second is below a floor value.  In the
example below the floor values are:

   input1.mnc 12.5
   input2.mnc 23.5

   minccalc -expression "(A[1] < 23.5) ? 0 : A[0]" input1.mnc
input2.mnc input1-floor.mnc

   minccmp -floor 12.5 input1-floor.mnc input2.mnc


a


On 29 January 2016 at 10:05, Paul GRAVEL <pgravel at bic.mni.mcgill.ca> wrote:
> Many thanks Andrew!
>
> I gave it a try, and it looks promising!
> btw, is there a way to floor each volume with different values?
>
> Best,
>
> Paul
>
>
>
> On Fri, 29 Jan 2016, Andrew Janke wrote:
>
>> Hi Paul,
>>
>> Yes the output of minctracc should be able to help you. Just be aware
>> that if you are running minctracc within something else (nlpfit,
>> mritotal, etc) in which step sizes and other things change that the
>> correlation will also change.
>>
>> You might also be interested in minccmp,  I added it some time ago for
>> this express purpose.
>>
>>
>> a
>>
>>
>> On 29 January 2016 at 05:38, Paul GRAVEL <pgravel at bic.mni.mcgill.ca>
>> wrote:
>>>
>>> Dear All,
>>>
>>> Hope all is well!
>>>
>>> Is there a metric (or metrics) to test the goodness of co-registration
>>> with
>>> minctracc (would that be comparing the Initial objective function val vs.
>>> Final objective function value)?
>>>
>>> The reason for this question is that I am running a script that does 2
>>> passes of a PET to PET co-registration, but sometimes the second pass
>>> messes
>>> things up, especially when the first pass looks good.
>>> Thus I would like to use this metric to automatically check whether or
>>> not
>>> the second pass should be run at all.
>>> Of course, I would still visually inspect the end results ;-)!
>>>
>>> Thanks!
>>>
>>> Paul
>>>
>>>
>>> _______________________________________________
>>> MINC-users at bic.mni.mcgill.ca
>>> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
>>
>> _______________________________________________
>> MINC-users at bic.mni.mcgill.ca
>> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
>>
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users