Dear Anderson and other PALMers,
I've run into some stubborn and confusing errors during my PALM runs and was hoping you could help diagnose the source of these issues. In addition to this unfortunately lengthy question (because of the details I'm trying to provide), I have two related questions as well:
1) Is it possible to produce both NPC and non-NPC (i.e., unimodal) results in the same palm run? For example, could you have produced the separate area and thickness results in addition to their joint NPC analysis during the same run in the 2018 Joint Analysis paper? Could the separate area and thickness results have been produced in the same run so that -corrmod could be used? Relatedly, have you made the palm commands (options etc.) for the 2018 Joint Analysis paper available anywhere? No worries if not, but figured I'd ask.
2) Many of the logp maps produced by my palm runs are entirely yellow (i.e., the default loaded overlay thresholds for a particular map - min/mid/max - are all set to 0 in Freeview, oftentimes with no results depicted in the Threshold section of 'Configure Overlay'). Have you encountered this before/have suggestions for how I might avoid these entirely "significant" but not actually significant maps?
Main question:
Here's an example of one of the palm commands that resulted in these output and error logs (with line-breaks for legibility):
fsl_sub -l $logdir palm -designperinput
-i {area_dir}/292_lh.area_stack.mgh
-d {design_dir}/wordadhd_npc_areatotal.mat
-t {design_dir}/wordadhd_npc_areatotal.con
-i {thickness_dir}/292_lh.thickness_stack.mgh
-d {design_dir}/wordadhd_npc_thickmean.mat
-t {design_dir}/wordadhd_npc_thickmean.con
-m {base_dir}/palm/input/$lhmask.mgz
-s ${surf_dir}/lh.white ${surf_dir}/lh.white.avg.area.mgh
-o wordadhd_npc_lh
-eb ${EB}
-corrcon -corrmod -logp -ise -npc -T -C 1.64485 -n 5 -nouncorrected
The error log contains the following text:
{^HCell contents reference from a non-cell array object.
Error in palm_saveall (line 1043)
distmax(:,j) = plm.Tclumax{m}{c};
Error in palm_core (line 2489)
palm_saveall(plm,opts);
Error in palm (line 81)
palm_core(varargin{:});
}^H
^G
Some more extensive details:
I'm analyzing area and thickness and have four one-tailed contrasts of interest (e.g., C1+, C1-, C2+, C2-). In case it matters, I used mri_concat to combine subjects' surf/lh.fwhm15.fsaverage.mgh results into a 4d input file. Currently, I am only able to get the "output'_dpv_npc_fisher_cfwep_c[1234]" results, but the palm log ends abruptly after saving these files and does not save the _tfce_ or _clustere_ versions:
Saving p-values for NPC between modalities (corrected across contrasts).
Saving file: wordadhd_npc_lh_dpv_npc_fisher_cfwep_c1
Saving file: wordadhd_npc_lh_dpv_npc_fisher_cfwep_c2
Saving file: wordadhd_npc_lh_dpv_npc_fisher_cfwep_c3
Saving file: wordadhd_npc_lh_dpv_npc_fisher_cfwep_c4
>>
I have essentially permuted the various -corr(con/mod) and -npc(con/mod) options, as well as various combinations of -T & -C {z}, -Tuni & -Cuni {z}, and -Tnpc & -Cnpc {z}, using a small number of permutations to assess the number of output maps and whether jobs finish appropriately. I've also reduced the number of contrasts to only run NPC for one result at a time (e.g., C1+, C1-, C2+, & C2- separately), and attempted other other acceleration and NPC (i.e., non-Fisher) methods. My attempts to read through the FreeSurfer and FSL forums, as well as the relevant source code on GitHub, haven't yet helped me produce my intended results.
If knowing versions helps at all:
PALM alpha109
MATLAB 8.3.0.532 (R2014a)
FreeSurfer 5.3.0
FSL 5.0.8
Any advice you could provide on this particular error, why my input doesn't produce the tfce_cfwep or clustere_cfwep results, or other problem-solving suggestions would be greatly appreciated!
With gratitude,
Dan
P.S. As many other folks have said on this and other forums, thank you for such a useful, versatile, and well-documented tool with accompanying literature!
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|