Thanks John,
I have tried changing the line 274 of spm_image.m to "if sum(tmp(:).^2)>1e-4," however this didn't make any difference and I still get the warning: shears involved when I display an image.
Using DICOM import I get QFORM0 representation has been rounded
Then I get the following:
N.mat
-0.7804 0.0201 0.1000 97.9979
-0.0300 0.0046 -1.6000 107.9834
0.0203 0.7810 0 -115.5794
0 0 0 1.0000
N.mat0
-0.7800 0.0202 0.0809 98.0166
-0.0394 0.0029 -1.6011 107.9955
0.0203 0.7810 0.0039 -115.5833
0 0 0 1.0000
Do you think this ok? Or would you advise that I covert images using a different package as I will be using DARTEL after all the within subject pre-processing.
Thanks in advance
Emma
>-----Original Message-----
>From: John Ashburner [mailto:[log in to unmask]]
>Sent: 27 October 2010 15:41
>To: e j burton
>Cc: [log in to unmask]
>Subject: Re: [SPM] Warning shears involved
>
>Whether it effects your data may depend on where exactly the
>error comes from. I would suggest checking some of the NIfTI
>images to make sure the converter has included the right DICOM
>files into each of the images. If it has, then the warning
>probably comes from some small rounding errors. Try changing
>line 274 of spm_image.m from "if sum(tmp(:).^2)>1e-8," to "if
>sum(tmp(:).^2)>1e-4,". This may make the problem go away.
>
>If the orientation of the data does include some shearing (as
>is common for CT data), most of the SPM processing should be
>OK - except for some features of Dartel. To get an idea of
>the magnitude of any problem, you could try following:
>
>N=nifti('my_image.img');
>N.mat
>N.mat0
>
>If the N.mat and N.mat0 matrices are very close, then there
>should be no problem. If they differ, it may mean that the
>fields that make up mat0 are unable to represent the shears.
>mat0 only encodes a 9-parameter affine transform
>(translations, voxel sizes and rotations), whereas mat encodes
>a full 12-parameter affine transform.
>Usually these are the same because the voxel-to-world matrices
>in the headers can be represented by 9 parameters. However,
>if there is some sort of shearing in the data, then 12
>parameters are needed, which mat0 can not encode.
>
>Best regards,
>-John
>
>
>On 27 October 2010 11:35, e j burton
><[log in to unmask]> wrote:
>> Dear SPMers
>>
>> When using SPM to convert the images from DICOM and then
>display them I get a warning: shears involved. I am wanting to
>do longituidnal analysis with the data and wonder if this will
>lead to errors in spatial normalisation?
>>
>> Should I worry about this?
>> Would it be best if the dicom images are converted using,
>for example MRIcro, prior to using SPM?
>>
>> I also need to reorient the images to the AC will this
>warning affect this? And when I alter pitch, roll or yaw, why
>does the image in the display appear smaller? And will this
>affect my preprocessing involving HDW, DARTEL etc?
>>
>> Thanks in advance for any help.
>>
>> Emma
>>
>
|