Hi Mayank, I may have a few more general considerations before commenting on the number and kind of random effects.
In the current design matrix, column 2 (time from baseline) and column 5 (visit) will be quite redundant. It is one of the advantages of the LME model to be able to deal with unevenly spaced observation intervals and also with possibly missing visits. In that sense, all temporal information is already contained within column 2, and I cannot see a good reason to additionally include column 5. It may even complicate things, due to the redundancy. If you have 3 groups, you will need two columns in the design matrix to indicate group membership (assuming that an intercept column is also present). Choose one group (say, group 1) as the reference group, which will not explicitly be coded, and then use one of the two group columns to indicate if a person belongs to group 2 (1=yes, 0=no), and the other to indicate if a person belongs to group 3 (1=yes, 0=no). Please don't code group membership numerically as 1, 2, 3 in a single column, as this will impose a linear relation between group membership and outcome measure (similar to linear regression). I am not sure about the current meaning of column 6, which multiplies age and time-from-baseline. It would be more important, in my eyes, to multiply time-from-baseline and group membership. Considering my second comment, this will also result in two colums. These will indicate whether or not the effect of time will differ between groups, and this (=different longitudinal trajectory) is often the primary interest in longitudinal studies. As far as I can see, your overall design is somewhat similar to the univariate example on the Freesurfer LME webpage, except that you have fewer groups and fewer covariates. As it does not matter for the design matrix whether or not the overall study is univariate or mass- univariate (as in your case), one suggestion would be that you construct your design in a similar way as in the given example. I hope that I have understood your design correctly, otherwise let me know. Best regards, Kersten -----Original Message----- From: "Kaushal, Mayank" <mkaus...@mcw.edu> Reply-to: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu> To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu> Subject: Re: [Freesurfer] Running Linear Mixed Effects (LME) Model Date: Tue, 6 Mar 2018 19:22:57 +0100 I have tabulated the columns for easier comprehension. Column 1 Column 2 Column 3 Column 4 Column 5 Column 6 intercept time of each scan from base scan for a given subject age at first scan for a given subject group (1, 2, or 3) visit (1,2 3, or 4) Column 2 * Column 3 Thanks again. Mayank On Mar 6, 2018, at 12:05 PM, Kaushal, Mayank <mkaus...@mcw.edu<mailto:m kaus...@mcw.edu>> wrote: Changing the Matlab directory worked. Thanks for all your inputs, Kersten. Sorry to prolong the conversation but if possible, I would appreciate input on whether my thought process regarding the LME analysis is on the right track. I am looking to compare cortical thickness between 3 groups over multiple visits. The design matrix created by me X = [ones(length(M),1) M M(:,1).*M(:,2)]; Column 1 Column 2 Column 3 Column 4 Column 5 Column 6 intercept time of each scan from base scan age at first scan group (1, 2, or 3) visit (1,2 3, or 4) Column 2 * Column 3 Presently, I am running lhstats = lme_mass_fit_vw(X,[1 2],Y,ni,lhcortex,[],1) as a test run. However, I would like to include Columns 1, 4, and 5 as random effects for LME analysis and if possible even include Column 2 as a random effect to have 4 random effects. lhstats = lme_mass_fit_vw(X,[1 4 5],Y,ni,lhcortex,[],1) or lhstats = lme_mass_fit_vw(X,[1 2 4 5],Y,ni,lhcortex,[],1) Now, I would like to know if having 3 or possibly 4 random effects would be too many random effects. Mayank On Mar 6, 2018, at 8:57 AM, Diers, Kersten /DZNE <kersten.di...@dzne.de <mailto:kersten.di...@dzne.de><mailto:kersten.di...@dzne.de>> wrote: ATTENTION: This email originated from a sender outside of MCW. Use caution when clicking on links or opening attachments. ________________________________ The LME toolbox will create a text file ('parfor_progress1.txt') in Matlab's present working directory to indicate its progress in processing each vertex. To create this file it needs to have write permissions in Matlab's present working directory. I speculate that your present working directory in Matlab is the root directory '/' where the toolbox cannot create a file. The solution would be to not run the scripts from '/', but to set the Matlab working directory to another directory where you have write permissions. Best regards Kersten -----Original Message----- From: "Kaushal, Mayank" <mkaus...@mcw.edu<mailto:mkaus...@mcw.edu><mail to:mkaus...@mcw.edu>> Reply-to: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu<mailt o:freesurfer@nmr.mgh.harvard.edu>> To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu<mailto:free sur...@nmr.mgh.harvard.edu>> Subject: Re: [Freesurfer] Running Linear Mixed Effects (LME) Model Date: Tue, 6 Mar 2018 15:17:31 +0100 After downloading version 6.0. and running lhstats = lme_mass_fit_vw(X,[1 2],Y,ni,lhcortex,[],1), I received the following errors. Error using parfor_progress (line 17) Do you have write permissions for /? Error in lme_mass_fit (line 155) fn = parfor_progress('init',nv); Error in lme_mass_fit_vw (line 73) [stats1,st1] = lme_mass_fit(X,[],Xrows,Zcols,Y,ni,prs,e); Thoughts? Mayank On Mar 6, 2018, at 7:08 AM, Diers, Kersten /DZNE <Kersten.Diers@dzne. de> wrote: ATTENTION: This email originated from a sender outside of MCW. Use caution when clicking on links or opening attachments. ________________________________ Yes, my understanding is the same. The LME toolbox can in a way be regarded as a post-processing add-on to Freesurfer. The only interface with Freesurfer are the read/write functions, everything else is to the best of my knowledge independent of Freesurfer. So I cannot see any problems in using the newer version of the LME toolbox with data from Freesurfer 5.3. Needless to say, this does not affect the usual recommendation, which is that Freesurfer versions should not be mixed in general. Best, Kersten ------Original Message----- From: "Kaushal, Mayank" <mkaus...@mcw.edu<mailto:mkaus...@mcw.edu>> Reply-to: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu<mailt o:freesurfer@nmr.mgh.harvard.edu>> To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu<mailto:free sur...@nmr.mgh.harvard.edu>> Subject: Re: [Freesurfer] Running Linear Mixed Effects (LME) Model Date: Mon, 5 Mar 2018 20:47:36 +0100 The version string I have been using: $Revision: 1.1.1.1 $ $Date: 2012/02/02 11:25:52 $ So, it appears that I do indeed have the older version. I am running Matlab 2017a, which is a relatively newer version. Before I proceed with using LME toolbox from a freesurfer 6.0, I would appreciate your input on the following. My understanding is that LME toolbox is used for statistical analysis and is independent of the freesurfer version performing the actual cortical thickness analysis. So, in theory, if I import cortical thickness data processed in version 5.3, I should still be able to run the LME toolbox from a version 6.0. So, can I input the files processed by freesurfer 5.3 (files that have been processed till mris_preproc and mri_surf2surf) and simply use the LME toolbox from a version 6.0 to perform the statistical analysis? Mayank On Mar 5, 2018, at 1:29 PM, Diers, Kersten /DZNE <Kersten.Diers@dzn e. de> wrote: ATTENTION: This email originated from a sender outside of MCW. Use caution when clicking on links or opening attachments. ________________________________ Hi, thanks for the additional information. I believe that we are dealing with a version issue now. Specifically, I have so far been testing with an LME toolbox from a Freesurfer 6.0 version on Matlab R2016a. In that sense, there is no need for a separate download of the LME scripts, since they are already included within the freesurfer directory (e.g. 'freesurfer/matlab/lme'). It is primarily the 'lme_mass_fit.m' script, which throws the current error. Could you please check your version of this script (type 'help lme_mass_fit.m')? Within my 5.3 installation, the version string is: % $Revision: 1.1.2.2 $ $Date: 2013/02/23 21:08:10 $ Within my 6.0 installation, the version string is: $Revision: 1.3 $ $Date: 2016/04/08 19:39:24 $ Unfortunately, the version string of the 'lme_mass_fit.m' script within a toolbox that I just downloaded from the LME website is the most ancient one: % $Revision: 1.1.1.1 $ $Date: 2012/02/02 11:25:52 $ When I use the 1.1.1.1 version instead of the one that is included within Freesurfer 6.0, I can replicate the problem reported in your current mail. With the 1.3 version I have no issues. As far as I know, the 'lme_mass_fit.m' script was updated to accommodate the changes in Matlab's way of doing the parallel processing. In older Matlab versions this was done using the 'matlabpool' command, but this was replaced by 'parpool', which is - to my knowledge - the only way of running parallel processing with recent Matlab versions. The LME toolbox was originally only tailored to the 'matlabpool' command, but has later been adapted to also include the 'parpool' command. Assuming that you are using a relatively recent Matlab version, the use of earlier vs. later versions of the toolbox would explain our respective observations. All in all, I currently see no way of getting an older version of the LME toolbox to run with a current Matlab version - except from modifying the code itself. So my suggestion would be to just use the newer LME toolbox that is shipped with Freesurfer 6.0. Let us know if I am mistaken or if there are further issues. Best regards, Kersten -----Original Message----- From: "Kaushal, Mayank" <mkaus...@mcw.edu<mailto:mkaus...@mcw.edu>> Reply-to: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu<mailt o:freesurfer@nmr.mgh.harvard.edu>> To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu<mailto:free sur...@nmr.mgh.harvard.edu>> Subject: Re: [Freesurfer] Running Linear Mixed Effects (LME) Model Date: Mon, 5 Mar 2018 16:34:14 +0100 Thanks for the quick response Kersten. I tried all the potential suggestions mentioned in your post but still no luck. Here’s the summary: lhstats = lme_mass_fit_vw(X,[1 2],Y,ni,lhcortex,[],1) Undefined function or variable 'matlabpool'. Error in lme_mass_fit (line 124) if (matlabpool('size') > 0) Error in lme_mass_fit_vw (line 73) [stats1,st1] = lme_mass_fit(X,[],Xrows,Zcols,Y,ni,prs,e); lhstats = lme_mass_fit_vw(X,[1 2],Y,ni,lhcortex,[],2) Undefined function or variable 'matlabpool'. Error in lme_mass_fit (line 123) if (prs==1) || (matlabpool('size') ~= prs) Error in lme_mass_fit_vw (line 73) [stats1,st1] = lme_mass_fit(X,[],Xrows,Zcols,Y,ni,prs,e); The "automatically create a parallel pool" was already deselected in my case. Also, I am using freesurfer 5.3 with LME toolbox downloaded from the freesurfer website. Mayank On Mar 5, 2018, at 9:12 AM, Diers, Kersten /DZNE <Kersten.Diers@d zn e. de> wrote: ATTENTION: This email originated from a sender outside of MCW. Use caution when clicking on links or opening attachments. ________________________________ Hi, you might try supplying an additional command-line argument while calling the 'lme_mass_fit_vw' function: lhstats = lme_mass_fit_vw(X,[1 2],Y,ni,lhcortex,[],1) This will tell Matlab to use only one worker. Further documentation on this additional argument can be obtained by typing 'help lme_mass_fit_vw' at the Matlab prompt. To get this working with my Matlab installation, I additionally had to deselect the "Automatically create parallel pool (if one doesn't already exist) when parallel keywords (e.g., parfor) are executed." option. This can be found in Matlab's preferences menu (--> see the Parallel Computing toolbox section). As an alternative, using two workers might also be an option in your case: lhstats = lme_mass_fit_vw(X,[1 2],Y,ni,lhcortex,[],2) To get this alternative approach working, I did not have to change the Matlab preferences. Hope this helps, Kersten -----Original Message----- From: "Kaushal, Mayank" <mkaus...@mcw.edu<mailto:mkaus...@mcw.edu>> Reply-to: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu<mailt o:freesurfer@nmr.mgh.harvard.edu> To: freesurfer@nmr.mgh.harvard.edu<mailto:freesur...@nmr.mgh.harvard.ed u> <freesur...@nmr.mgh.harvard.ed<mailto:freesur...@nmr.mgh.harvard.ed> u> Subject: [Freesurfer] Running Linear Mixed Effects (LME) Model Date: Sun, 4 Mar 2018 23:56:03 +0100 Dear Freesurfer Community, Running lhstats = lme_mass_fit_vw(X,[1 2],Y,ni,lhcortex); gives the following error, Starting parallel pool (parpool) using the 'local' profile ... Error using parpool (line 104) You requested a minimum of 8 workers, but the cluster "local" has the NumWorkers property set to allow a maximum of 2 workers. To run a communicating job on more workers than this (up to a maximum of 512 for the Local cluster), increase the value of the NumWorkers property for the cluster. The default value of NumWorkers for a Local cluster is the number of cores on the local machine. Error in lme_mass_fit (line 137) parpool(prs); Error in lme_mass_fit_vw (line 73) [stats1,st1] = lme_mass_fit(X,[],Xrows,Zcols,Y,ni,prs,e); I am running Matlab on a personal MacBook Pro, 2.4 GHz Intel Core i5, 8 GB 1600 MHz DDR3. I have researched the forum on how to stop the function from trying to initiate parallel processing but so far haven’t been able figure out how to overcome this. Any help would be much appreciated. Mayank Kaushal MD MBA Postdoctoral Fellow Department of Neurosurgery Medical College of Wisconsin _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu> https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh .h ar vard.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=aFamLAsxMIDYjNglY HT MV 0iqFn3z4pVFYPQkjgspw4Y&r=RqvEwdmcEg_aWsE7PBL80w&m=o0oLymaypGqFbkf AR - cOnHzq1lBIA47IRaf9HlVhH8c&s=JDoMa8C- 2GQ4TpMIyRj4xZ0XosQFEMbzwvKLY9s5G9M&e= The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners. or g_ complianceline&d=DwIGaQ&c=aFamLAsxMIDYjNglYHTMV0iqFn3z4pVFYPQkjgs pw 4Y &r=RqvEwdmcEg_aWsE7PBL80w&m=o0oLymaypGqFbkfAR- cOnHzq1lBIA47IRaf9HlVhH8c&s=4ORZf53F92dPjiBGXnKHpr- EpQ0EcHphHdwjrXBmi2M&e= . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu> https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh .h ar vard.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=aFamLAsxMIDYjNglY HT MV 0iqFn3z4pVFYPQkjgspw4Y&r=RqvEwdmcEg_aWsE7PBL80w&m=o0oLymaypGqFbkf AR - cOnHzq1lBIA47IRaf9HlVhH8c&s=JDoMa8C- 2GQ4TpMIyRj4xZ0XosQFEMbzwvKLY9s5G9M&e= _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu> https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.h ar vard.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=aFamLAsxMIDYjNglYHT MV 0iqFn3z4pVFYPQkjgspw4Y&r=RqvEwdmcEg_aWsE7PBL80w&m=OYRCgY8BRpqstbCR3 wn uPmKgzUIsDM1jpSeXZ5Vq_k4&s=vZuOCZdUz6e6UbqOzq6Wt- CfJpE_fTmXKmRXAEOva_U&e= The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.or g_ complianceline&d=DwIGaQ&c=aFamLAsxMIDYjNglYHTMV0iqFn3z4pVFYPQkjgspw 4Y &r=RqvEwdmcEg_aWsE7PBL80w&m=OYRCgY8BRpqstbCR3wnuPmKgzUIsDM1jpSeXZ5V q_ k4&s=l79g5eOp0ctiFMn0i9TPGz-xgQD77NyKMbK1SskWckA&e= . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu> https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.h ar vard.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=aFamLAsxMIDYjNglYHT MV 0iqFn3z4pVFYPQkjgspw4Y&r=RqvEwdmcEg_aWsE7PBL80w&m=OYRCgY8BRpqstbCR3 wn uPmKgzUIsDM1jpSeXZ5Vq_k4&s=vZuOCZdUz6e6UbqOzq6Wt- CfJpE_fTmXKmRXAEOva_U&e= _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu> https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.har vard.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=aFamLAsxMIDYjNglYHTMV 0iqFn3z4pVFYPQkjgspw4Y&r=RqvEwdmcEg_aWsE7PBL80w&m=1p97jFA3j0xKIWnKfnu 9m0d_7s53UXt0ZNl0blCKxNs&s=hnOncrG2_ylJ3FodlXAbE1EVVhUUtXgGEnkiGlXXh8 c&e= The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_ complianceline&d=DwIGaQ&c=aFamLAsxMIDYjNglYHTMV0iqFn3z4pVFYPQkjgspw4Y &r=RqvEwdmcEg_aWsE7PBL80w&m=1p97jFA3j0xKIWnKfnu9m0d_7s53UXt0ZNl0blCKx Ns&s=GexmzGtdEYAmFIlm6hCmI2663Z-CgUy8UHaLR-K7C2M&e= . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu> https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.har vard.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=aFamLAsxMIDYjNglYHTMV 0iqFn3z4pVFYPQkjgspw4Y&r=RqvEwdmcEg_aWsE7PBL80w&m=1p97jFA3j0xKIWnKfnu 9m0d_7s53UXt0ZNl0blCKxNs&s=hnOncrG2_ylJ3FodlXAbE1EVVhUUtXgGEnkiGlXXh8 c&e= _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu> https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=aFamLAsxMIDYjNglYHTMV0iqF n3z4pVFYPQkjgspw4Y&r=RqvEwdmcEg_aWsE7PBL80w&m=2vVWX9l8uCW6PNDofp_jxzkt5 1TCVmkFxY2vY-Ag_iM&s=GO_IQm-SiVRVkbf2CjC0vitbE36m5J11HMxymoSucXI&e= The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=aFamLAsxMIDYjNglYHTMV0iqFn3z4pVFYPQkjgspw4Y&r=R qvEwdmcEg_aWsE7PBL80w&m=2vVWX9l8uCW6PNDofp_jxzkt51TCVmkFxY2vY- Ag_iM&s=IjgRll5-M3Zti1umbPJ31L6xeflAsxz9YUcZS2jnsyE&e= . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu> https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=aFamLAsxMIDYjNglYHTMV0iqF n3z4pVFYPQkjgspw4Y&r=RqvEwdmcEg_aWsE7PBL80w&m=2vVWX9l8uCW6PNDofp_jxzkt5 1TCVmkFxY2vY-Ag_iM&s=GO_IQm-SiVRVkbf2CjC0vitbE36m5J11HMxymoSucXI&e= _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu> https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=aFamLAsxMIDYjNglYHTMV0iqF n3z4pVFYPQkjgspw4Y&r=RqvEwdmcEg_aWsE7PBL80w&m=JtHeh4F36Zz1ELE7LnYzZXcbl TZGl5bubrruLBAzfrw&s=N7bPtrlGX5UKHsj87HzubSjQ5VXnkSpEuYwo_I-ypJU&e= The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=aFamLAsxMIDYjNglYHTMV0iqFn3z4pVFYPQkjgspw4Y&r=R qvEwdmcEg_aWsE7PBL80w&m=JtHeh4F36Zz1ELE7LnYzZXcblTZGl5bubrruLBAzfrw&s=4 RDz2B3bkbGv045ePFVlaXOUhwU-3UOmzh11Xp8sYJc&e= . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
