[MINC-users] mritotal results depend on the filetype?

[Daniel Schwarz]

Yes, I use mincreshape for converting the file between the different 
data types.
I was very confused, because of big differencies between resulting 
transformations. The big differencies occured when the volume was 
cropped too much (I did some mistakes in cropping - my fault) and the 
registration process was therefore unsuccessful.
Now, my data are getting registered well. There are still some 
differencies in resulting transformations (depending on the data type), 
but there are very small and unimportant, I guess...

Daniel Schwarz


Andrew Janke napsal(a):

> On Fri, 22 Oct 2004, Daniel Schwarz wrote:
>
>> I am getting different results from mritotal when trying to register 
>> the same image. The results differs when I change data type from 
>> signed short to unsigned byte.
>> I tried it with XCORR and MI too. In the both cases I am getting 2 
>> different transformations for byte and int inputs (2 different 
>> results for the same objective function and the same input image, 
>> only the depths are different.). Isn't it a bug?
>> In the case of INT data, mritotal is displaying info "source volume 
>> not UNSIGNED_BYTE, will do conversion now"....so why there are 
>> different results if I do the conversion before the registration 
>> process starts?
>>
>> I use mni_autoreg-0.98r (I got different results with 0.98d too).
>> My images are short int with range 0..4095.
>> I use default model average_305.mnc and I haven't changed anything in 
>> the mritotal.
>
>
> Daniel,
>
> Without looking at the data, one of the guesses I would be making is 
> that the threshold seelction is getting confused with the changes in 
> datatype.  How are you converting the file between the different data 
> types?  with mincreshape?
>
>
> -- 
> Andrew Janke (andrew_janke@iinet.net DOT au || 
> www.cmr.uq.edu.au/~rotor) Australia->Brisbane            H: +61 7 3390 
> 6332  || M: +61 4 2138 8581