Hi Anderson
Durning waiting your kindly response, we have tried to run the PALM with
"-accel gamma" and this time everything run smoothly without any error. I
think the acceleration methods which you developed are quiet helpful in
this "relative large sample size" study and we wish we could still use
these methods in this study. In our original study design, we ran
cerebellum and cerebral cortex VBM seperately. I will upload the part of
preprocessed datset to the link you provide when I go to the office.
In addtion, do PALM output the cluster extent FDR corrected map ? I only
obtain TFCE FWE&FDR corrected P map after running PALM.
Best
Paul
2016-09-06 14:52 GMT+08:00 Anderson M. Winkler <[log in to unmask]>:
> Hi Paul,
>
> Thanks for sending. Design and contrasts look fine. Perhaps it's an odd
> combination of values in the data (and perhaps with TFCE). If you could
> upload your data I would be happy to have a look. You should have just
> received an automated OxFile link for the upload. Please also include the
> statistic.fts file (i.e., the F-test).
>
> In any case, for this design and these options, except for the "-accel
> tail" (which where the error is arising), this analysis can be done in
> randomise, so maybe consider using it regardless of what the problem might
> be or when we will be able to figure it out. Thanks.
>
> All the best,
>
> Anderson
>
> PS: sorry for the typo in your name, I just noticed after having sent the
> link.
>
>
> On 5 September 2016 at 08:29, KUN HSIEN CHOU <[log in to unmask]>
> wrote:
>
>> Dear Anderson
>>
>> The following attached files are contrast and design file for the current
>> study. Due to the large sample size of the current study, it is very slow
>> when we try to paste the design into GLM gui, so we use text editor to
>> creat these two file (Maybe there are some error usage when we create these
>> two files).
>>
>> PS: For the statistic.mat. The first three colume are group lable (we
>> have three study groups in this study). The last two colume are demean age
>> and sex.
>>
>> Thanks very much.
>>
>>
>>
>> 2016-09-05 14:53 GMT+08:00 Anderson M. Winkler <[log in to unmask]>:
>>
>>> Hi Paul, hi Niels,
>>>
>>> PALM removes the NaN and Infs before starting so this shouldn't be the
>>> issue. However, depending on the values in the data, design and/or
>>> contrasts, new NaNs or Infs can be created during the analysis, leading to
>>> that error. One possibility is that the test is assessing whether the group
>>> mean is different than zero, which is always the case for GM volume,
>>> leading to statistics that approach infinity, then the error.
>>>
>>> Paul: could you show your design and contrasts? Thanks.
>>>
>>> All the best,
>>>
>>> Anderson
>>>
>>>
>>> On 5 September 2016 at 03:13, KUN HSIEN CHOU <[log in to unmask]>
>>> wrote:
>>>
>>>> For s2_GM_1197.nii : 0.000000 2.378755
>>>> For mask.nii: 0.000000 1.000000
>>>>
>>>> Best
>>>>
>>>> Paul
>>>>
>>>> 2016-09-05 6:37 GMT+08:00 Niels Bergsland <[log in to unmask]>:
>>>>
>>>>> What do you get if you run fslstats -R on your input image and mask?
>>>>>
>>>>>
>>>>> On Monday, September 5, 2016, KUN HSIEN CHOU <[log in to unmask]>
>>>>> wrote:
>>>>>
>>>>>> Hi
>>>>>>
>>>>>> The input data file and the corresponded mask file were also run
>>>>>> through fslmaths with the -nan option before entering into the PALM
>>>>>> software , but the error still occured.
>>>>>>
>>>>>> Best
>>>>>>
>>>>>> Paul
>>>>>>
>>>>>> 2016-09-05 1:05 GMT+08:00 Niels Bergsland <[log in to unmask]>:
>>>>>>
>>>>>>> Hi. I haven't had this issue myself but it sounds like you might
>>>>>>> need to run your input image through fslmaths with the -nan option.
>>>>>>>
>>>>>>> On Sun, Sep 4, 2016 at 4:57 PM, Paul Chou <[log in to unmask]>
>>>>>>> wrote:
>>>>>>>
>>>>>>>> Dear PALM developers
>>>>>>>>
>>>>>>>> I tried to perform permutation test for our VBM dataset (we use SPM
>>>>>>>> for VBM preprocessing) using PALM software. After running permutation test,
>>>>>>>> the error message comes out in the MATLAB command window. Could you help me
>>>>>>>> to deal with this error message ? In addtion, although the error message
>>>>>>>> occured, PALM still produced s2_ILAS_1197_vox_tstat.nii for each
>>>>>>>> statistical contrast.
>>>>>>>>
>>>>>>>> I use the following command for PALM analysis (We use the latest
>>>>>>>> version of PALM and Matlab R2012a.). The s2_GM_1197.nii contains 1197
>>>>>>>> subjects VBM preprocessed dataset.
>>>>>>>>
>>>>>>>> palm -i /mnt/HD2T/PLAM_statistic/S2_GM_1197/s2_GM_1197.nii -m
>>>>>>>> dartel4_mask.nii -d statistic.mat -t statistic.con -f statistic.fts -T -C
>>>>>>>> 2.58 -n 1500 -fdr -save1-p -accel tail -o /mnt/HD2T/PLAM_statistic/S2_GM
>>>>>>>> _1197/s2_ILAS_1197
>>>>>>>>
>>>>>>>> The following message is error message :
>>>>>>>> =======================================================
>>>>>>>> Computing p-values.
>>>>>>>> Error using griddedInterpolant
>>>>>>>> The coordinates of the input points must be finite values; Inf and
>>>>>>>> NaN are not permitted.
>>>>>>>>
>>>>>>>> Error in interp1>Interp1D (line 335)
>>>>>>>> F = griddedInterpolant(X,V(:,1),method);
>>>>>>>>
>>>>>>>> Error in interp1 (line 220)
>>>>>>>> Vq = Interp1D(X,V,Xq,method);
>>>>>>>>
>>>>>>>> Error in palm_pareto>andersondarling (line 195)
>>>>>>>> i2 = interp1(i1,ptable,A2,'linear','extrap');
>>>>>>>>
>>>>>>>> Error in palm_pareto (line 127)
>>>>>>>> A2pval = andersondarling(gpdpvals(ytail,apar,kpar),kpar);
>>>>>>>>
>>>>>>>> Error in palm_saveall (line 77)
>>>>>>>> plm.Gpperm{y}{m}{c}(1,t) =
>>>>>>>> palm_pareto(plm.G{y}{m}{c}(1,t
>>>>>>>> ),plm.Gperms{y}{m}{c}(:,t),false,opts.accel.tail_thr,opts.ac
>>>>>>>> cel.G1out);
>>>>>>>>
>>>>>>>> Error in palm_core (line 2476)
>>>>>>>> palm_saveall(plm,opts);
>>>>>>>>
>>>>>>>> Error in palm (line 81)
>>>>>>>> palm_core(varargin{:});
>>>>>>>> ==================================
>>>>>>>>
>>>>>>>> Thank you
>>>>>>>>
>>>>>>>> Paul
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>>> Niels Bergsland
>>>>>>> Integration Director
>>>>>>> Buffalo Neuroimaging Analysis Center
>>>>>>> 100 High St. Buffalo NY 14203
>>>>>>> [log in to unmask]
>>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>> --
>>>>> Niels Bergsland
>>>>> Integration Director
>>>>> Buffalo Neuroimaging Analysis Center
>>>>> 100 High St. Buffalo NY 14203
>>>>> [log in to unmask]
>>>>>
>>>>
>>>>
>>>
>>
>
