Matlab version dependency

[tsieger]

Hi,
this is just a friendly notice that the current Lead-dbs v2.2.3 requires Matlab R2016b and higher, because it makes use of the contains() function, which was introduced in R2016b. This is in contrast to Lead-dbs installation instructions and prerequisites that declare that Lead-dbs requires Matlab 2015b and >R1014b, respectively.

In an attempt to make Lead-dbs 2.2.3 run on Matlab R2016a, I introduced a quick-and-dirty hacked version of contains.m like this:

function c=contains(a,b)
    c=~isempty(strfind(a,b));
end

This workarounds the issue of undefined contains(), but Lead-dbs does not seem to work properly, even though I do not know whether it is related to the contains issue. E.g., I ended up in debug mode in the ea_normalize_spmshoot function (see the keyboard keyword) when trying to normalize:

% check if darteltemplate is available, if not generate one
if exist([ea_space(options,'dartel'),'shootmni_6.nii'],'file')
    % There is a DARTEL-Template. Check if it will match:
    Vt=spm_vol([ea_space(options,'dartel'),'shootmni_6.nii']);
    Vp=spm_vol([directory,'rc1',options.prefs.prenii_unnormalized]);
    if ~isequal(Vp.dim,Vt(1).dim) || ~isequal(Vp.mat,Vt(1).mat) % Dartel template not matching. -> create matching one.
        ea_create_tpm_darteltemplate; %([directory,'rc1',options.prefs.prenii_unnormalized]);
    end
else % no dartel template present. -> Create matching dartel templates from highres version.
    keyboard
    ea_create_tpm_darteltemplate; %([directory,'rc1',options.prefs.prenii_unnormalized]);
end

Best wishes,
Tomas

[andreashorn]

Hi Thomas,

Great, thanks for catching this one. Would you like to submit a pull request that fixes the contains issue?
Re shoot haven’t used it in a while, would recommend to use ANTs.
But can have a look at it if I don’t forget.

Best, Andy

[andreashorn]

…based on the code, the keyboard should be deleted, so you can just run the next line to create TPMs.

[tsieger]

Regarding the contains() issue: I’m afraid my hack was too hacky and could not be used as a proper replacement for the original function, as it e.g. does not support patterns consisting of several strings, which the original function does support. Also, contains() can be used in both case-sensitive and insensitive modes, while strfind cannot. However, looking in the Lead-DBS code, multiple-string patterns seem not to be used, and possibly neither case-insensitive search.
So, I suppose there are three possible solutions:
i) put the hacked version of contains() in the code *only* for Matlab <R12016b (I can’t see a way how to do this when there is no installation process and users can get the code directly from github),
ii) put the hacked version of contains() in the code regardless of Matlab version (this would IMHO produce tons of warnings, I would not go this way),
iii) put a differently named function (e.g. mycontains() in the code and update all the calls to contains() to mycontains() – this would affect quite many people and much code, but would be quite clean IMHO.
What’s your opinion?

Regarding the normalization issue: ignoring the keyboard did not help, the process failed with

------------------------------------------------------------------------
Running job #1
------------------------------------------------------------------------
Running 'Run Shoot (existing Templates)'
Failed  'Run Shoot (existing Templates)'
Error using read_hdr (line 32)
Error reading header file "/home/sieger/src/matlab/leaddbs-2.2.3/templates/space/MNI_ICBM_2009b_NLIN_ASYM/dartel/shootmni_1.nii".
In file "/home/sieger/src/matlab/spm12/@nifti/private/read_hdr.m" (v4967), function "read_hdr" at line 32.
In file "/home/sieger/src/matlab/spm12/@nifti/nifti.m" (v4986), function "nifti" at line 26.
In file "/mnt/hdd/home/sieger/src/matlab/spm12/toolbox/Shoot/spm_shoot_warp.m" (v6024), function "load_template" at line 195.
In file "/mnt/hdd/home/sieger/src/matlab/spm12/toolbox/Shoot/spm_shoot_warp.m" (v6024), function "spm_shoot_warp" at line 104.

The following modules did not run:
Failed: Run Shoot (existing Templates)

Error using MATLABbatch system
Job execution failed. The full log of this run can be found in MATLAB command window, starting with the lines (look for the line showing the exact #job as displayed in this error message)
------------------
Running job #1
------------------

The problem was that the .../MNI_ICBM_2009b_NLIN_ASYM/dartel/shootmni_1.nii is missing.
There are only these files in the directory:

dartelmni_6.mat
dartelmni_6.nii
dartelmni_6_hires_00001.nii
dartelmni_6_hires_00002.nii
dartelmni_6_hires_00003.nii
dartelmni_6_hires_00004.nii
dartelmni_6_hires_00005.nii
dartelmni_6_hires_00006.nii
s0dartelmni_6_hires_00001.nii
s0dartelmni_6_hires_00002.nii
s0dartelmni_6_hires_00003.nii

[andreashorn]

Hi Tomasz,

re contains: Would use a function called ea_contains and replace the calls to that. Should never name a function the same as built-in matlab functions. There aren’t many calls of contains in Lead-DBS.

Re Shoot, not sure, cannot debug right now. However wouldn’t really recommend shoot for DBS based on the Ewert 2019 Neuroimage article. Is there a particular reason you have to use it?

Can look into it somewhen but currently am on holidays.

Best, Andy

[andreashorn]

…one thing you could try re Shoot is to run DARTEL (which also builds the TPMs if they are not present). If that works, afterwards, Shoot should also work.
Have asked Ningfei & Simon to look into this, maybe one of them has some time for it.

Best, Andy

[tsieger]

Re contains: I’ve issued this PR: https://github.com/netstim/leaddbs/pull/320

Re Shoot: I had no specific reason to use it, but it seems that ‘SPM12 Shoot’ is the default normalization option.

[ningfei]

Hi Tomas,

From the contents in your MNI_ICBM_2009b_NLIN_ASYM/dartel folder as you showed, it seems the function ea_create_tpm_darteltemplate (which will generate all the dartel and shoot files) didn’t run through. Maybe you can delete the folder MNI_ICBM_2009b_NLIN_ASYM/dartel and the file MNI_ICBM_2009b_NLIN_ASYM/TPM.nii, and run ea_create_tpm_darteltemplate again from MATLAB command window. You can then send me the full log. I will check out what’s happening during the template generation.

Best,
Ningfei

[andreashorn]

Great, thanks for the PR

[tsieger]

Hi Ningfei,
I tracked the issue down. This is also a Matlab dependency problem. The problem is in ea_delete.m, which makes use of the .folder field in the return value from the dir function.
In the current Matlab 2019b, the entries returned contain the following fields:

   name
    folder
    date
    bytes
    isdir
    datenum

but in R2016a, they contain only:

    name
    date
    bytes
    isdir
    datenum

This problem demonstrates in ea_create_tpm_darteltemplate, where a loop is made (l.147-194), cleaning using ea_delete on each iteration, and crashing at the end of the first iteration, never reaching the others, such that not all templates get created.

[tsieger]

Hi, I ran into another Matlab dependency problem with the newline function and create a PR fix for it: https://github.com/netstim/leaddbs/pull/321.

[tsieger]

Another problem in interactive 3d viewer control:

Error in ea_mouse_camera/motion_callback (line 63)
        pt = matlab.graphics.interaction.internal.getPointInPixels(hfig,currpt(1:2)); 
Error while evaluating Figure WindowButtonMotionFcn

this unfortunatelly also does not work on R2016a :-(.

[tsieger]

Another problem in vecnorm, which was introduced in R2017b.
https://github.com/netstim/leaddbs/pull/322

Please note that this PR also fixed the problem with incompatible matrix sizes:

Error using -
Matrix dimensions must agree.
Error in ea_genvat_horn>jr_remove_electrode (line 678)
tra = elstruct.trajectory{1,side} - org;
Error in ea_genvat_horn (line 355)
vat = jr_remove_electrode(vat,elstruct,mesh,side,elspec);
Error in ea_stimparams>stimulate_Callback (line 950)
            [stimparams(1,side).VAT(el).VAT,volume]=feval(ea_genvat,elstruct(el).coords_mm,getappdata(handles.stimfig,'S'),side,options,stimname,options.prefs.machine.vatsettings.horn_ethresh,handles.stimfig);
Error in gui_mainfcn (line 95)
        feval(varargin{:});
Error in ea_stimparams (line 42)
    gui_mainfcn(gui_State, varargin{:});
Error in matlab.graphics.internal.figfile.FigFile/read>@(hObject,eventdata)ea_stimparams('stimulate_Callback',hObject,eventdata,guidata(hObject)) 
Error while evaluating UIControl Callback

(I’ve used an explicit repmat to fix this.)

[tsieger]

However, now I run into a problem with:

Subscripted assignment dimension mismatch.
Error in ea_genvat_horn>jr_remove_electrode (line 712)
r_pos(:,r_pos(2,:)>0) = cr_pos;
Error in ea_genvat_horn (line 355)
vat = jr_remove_electrode(vat,elstruct,mesh,side,elspec);
Error in ea_stimparams>stimulate_Callback (line 950)
            [stimparams(1,side).VAT(el).VAT,volume]=feval(ea_genvat,elstruct(el).coords_mm,getappdata(handles.stimfig,'S'),side,options,stimname,options.prefs.machine.vatsettings.horn_ethresh,handles.stimfig);
Error in gui_mainfcn (line 95)
        feval(varargin{:});
Error in ea_stimparams (line 42)
    gui_mainfcn(gui_State, varargin{:});
Error in matlab.graphics.internal.figfile.FigFile/read>@(hObject,eventdata)ea_stimparams('stimulate_Callback',hObject,eventdata,guidata(hObject)) 
Error while evaluating UIControl Callback

size of rpos:
3 98658

size of r_pos(:,r_pos(2,:)>0):
3 89817

size of crpos:
3 179622

[Author]

Posts