Dear Mike, list,
I posted this issue a couple of times on the list. I can second that
this is caused by the determination of parameters of the autoregressive
modeling during the first (ReML) pass, and it does make sense that for
voxels containing only air the AR model developed for BOLD type temporal
correlations does not converge well.
We did not encounter this issue anymore when setting the masking
threshold back to anything signficantly larger than 0, it is 0.8 by
default when I remember well.
Thanks for sharing this with us, sounds like a good work-around.
Cheers,
Bas
Mike Wometz schreef:
> Hi,
>
> I've seen many postings on this spdiags error and experienced it first hand
> recently. It may result for a host of reasons, but one sure fire way to get
> this error is to use the intrinsic autocorrelation option (AR(1) + w) in
> your model in combination with a setting of the defaults.mask.thresh to 0.0
> in the spm_defaults. If you're using the mask threshold of 0.0 to get around
> excluding voxels in your second-level group analysis, but still want to use
> the intrinsic autocorrelation in your first-level, I believe you can
> increase your mask threshold for the first-level, then convert the NaNs in
> your con files to zeros, then decrease the mask threshold to 0.0 and run
> your second-level analysis.
>
> good luck,
> Mike Wolmetz
> Johns Hopkins University
> Department of Cognitive Science
>
> On Mon, 6 Nov 2006 15:24:19 -0800, Jan Gläscher <[log in to unmask]>
> wrote:
>
>
>> Hi Amit,
>>
>> I am just guessing here, but have you tried changing your BLAS version in
>> MATLAB? I recently came across a similar problem on my Linux box using
>> Matlab 7.1 (R14SP3), albeit not with the same error message, but also with
>> non-convergence in the ReML iterations. Perhaps there is the same problem
>> with Matlab 7.x under Windows.
>>
>> I change the BLAS version to libmkl_def.so and it work for my laptop with a
>> Pentium M processor. Maybe it'll work for you, too. Please consult this
>> message by Andreas Bruns and the link to a Mathworks README therein on how
>> to change the BLAS version:
>>
>> http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0605&L=SPM&P=R11778
>>
>> Good Luck,
>> Jan
>>
>> Amit Etkin wrote:
>>
>>> Hi all-
>>>
>>> I am running into an odd problem. I recently switched from a mac running
>>> unix matlab 6.5 and spm2 to a PC running windows matlab 7.0.1 and spm2. When
>>> setting up and estimating a first level model that has never given me
>>> problems before I run into the following problem on the PC (this is the end
>>> of the estimation from the matlab command window). The error is below, but
>>> the fun's not over yet...more problems further down the email:
>>>
>>> Plane 69/69 , block 14/23 : ...read & mask data
>>> Plane 69/69 , block 15/23 : ...read & mask data
>>> Plane 69/69 , block 16/23 : ...read & mask data
>>> Plane 69/69 , block 17/23 : ...read & mask data
>>> Plane 69/69 , block 18/23 : ...read & mask data
>>> Plane 69/69 , block 19/23 : ...read & mask data
>>> Plane 69/69 , block 20/23 : ...read & mask data
>>> Plane 69/69 , block 21/23 : ...read & mask data
>>> Plane 69/69 , block 22/23 : ...read & mask data
>>> Plane 69/69 , block 23/23 : ...done
>>> Temporal non-sphericity (over voxels) : ...REML estimation
>>> ReML Block - 1
>>> ReML Iteration : 1 ...NaN
>>> ReML Iteration : 2 ...NaN
>>> ReML Iteration : 3 ...NaN
>>> ReML Iteration : 4 ...NaN
>>> ReML Iteration : 5 ...NaN
>>> ReML Iteration : 6 ...NaN
>>> ReML Iteration : 7 ...NaN
>>> ReML Iteration : 8 ...NaN
>>> ReML Iteration : 9 ...NaN
>>> ReML Iteration : 10 ...NaN
>>> ReML Iteration : 11 ...NaN
>>> ReML Iteration : 12 ...NaN
>>> ReML Iteration : 13 ...NaN
>>> ReML Iteration : 14 ...NaN
>>> ReML Iteration : 15 ...NaN
>>> ReML Iteration : 16 ...NaN
>>> ReML Iteration : 17 ...NaN
>>> ReML Iteration : 18 ...NaN
>>> ReML Iteration : 19 ...NaN
>>> ReML Iteration : 20 ...NaN
>>> ReML Iteration : 21 ...NaN
>>> ReML Iteration : 22 ...NaN
>>> ReML Iteration : 23 ...NaN
>>> ReML Iteration : 24 ...NaN
>>> ReML Iteration : 25 ...NaN
>>> ReML Iteration : 26 ...NaN
>>> ReML Iteration : 27 ...NaN
>>> ReML Iteration : 28 ...NaN
>>> ReML Iteration : 29 ...NaN
>>> ReML Iteration : 30 ...NaN
>>> ReML Iteration : 31 ...NaN
>>> ReML Iteration : 32 ...NaN
>>>
>>> SPM2: spm_spm (v2.66) 09:21:00 - 06/11/2006
>>> ========================================================================
>>> Initialising parameters : ...computing???
>>> Index exceeds matrix dimensions.
>>>
>>> Error in ==> spdiags at 114
>>> a((len(k)+1):len(k+1),:) = [i i+d(k) B(i+(m>=n)*d(k),k)];
>>>
>>> Error in ==> spm_spm at 420
>>> s = spdiags(1./sqrt(diag(s)),0,nScan,nScan);
>>>
>>> Error in ==> spm_spm at 827
>>> SPM = spm_spm(SPM);
>>>
>>> ??? Error while evaluating uicontrol Callback.
>>>
>>>
>>>
>>> This problem has been posted on in the list before, but nobody has come up
>>> with a solution, though switching to different versions of matlab has been
>>> suggested. I then downloaded spm5 and tried running the same model with spm5
>>> and get the following error at the end of estimation:
>>>
>>> Plane 69/69 , block 14/23 : ...read & mask data
>>> Plane 69/69 , block 15/23 : ...read & mask data
>>> Plane 69/69 , block 16/23 : ...read & mask data
>>> Plane 69/69 , block 17/23 : ...read & mask data
>>> Plane 69/69 , block 18/23 : ...read & mask data
>>> Plane 69/69 , block 19/23 : ...read & mask data
>>> Plane 69/69 , block 20/23 : ...read & mask data
>>> Plane 69/69 , block 21/23 : ...read & mask data
>>> Plane 69/69 , block 22/23 : ...read & mask data
>>> Plane 69/69 , block 23/23 : ...done
>>> Temporal non-sphericity (over voxels) : ...REML estimation
>>> ReML Block - 1
>>> ReML Iteration : 1 ...4.796763e+004
>>> ReML Iteration : 2 ...3.059561e+004
>>> ReML Iteration : 3 ...2.632660e+002
>>> ReML Iteration : 4 ...4.751650e-001
>>> ReML Iteration : 5 ...1.323482e-004
>>>
>>> SPM: spm_spm (v$Rev: 649 $) 20:15:56 - 06/11/2006
>>> ========================================================================
>>> Initialising parameters : ...done
>>> Output images :
>>> ...initialisingError running job:
>>> Error using ==> feval
>>> Undefined command/function 'offset'.
>>> --------------------------
>>> Done.
>>>
>>>
>>>
>>> Anybody know what's going on and how to fix it?????
>>>
>>>
>>> much much much appreciated!!!
>>>
>>> Amit
>>>
>>>
>> --
>> Jan Gläscher, Ph.D. Caltech Brain Imaging Center
>> +1 (626) 395-4976 (office) Caltech, Broad Center, M/C 114-96
>> +1 (626) 395-2000 (fax) 1200 California Blvd
>> [log in to unmask] Pasadena, CA 91125
>> ========================================================================
>>
--
--------------------------------------------------
Dr. S.F.W. Neggers
Division of Brain Research
Rudolf Magnus Institute for Neuroscience
Utrecht University Medical Center
Visiting : Heidelberglaan 100, 3584 CX Utrecht
Room B.01.1.03
Mail : Huispost B.01.206, P.O. Box 85500
3508 GA Utrecht, the Netherlands
Tel : +31 (0)30 2509609
Fax : +31 (0)30 2505443
E-mail : [log in to unmask]
Web : http://www.fmri.nl/people/bas.html
--------------------------------------------------
|