JISCMail - FSL Archives

Dear FSL experts

I estimated spectral DCM, fully connected with 8 regions, and another spectral DCM with some connections off. I did it separately for all my subjects. After DCM estimations I used the following code. For example I choose only 2 subjects, so N=2

GCM = {'DCM_Subj1_full_modelFull_estimated.mat','DCM_Subj1_fullTimeSeries_modelAlternate_estimated.mat';

'DCM_Subj2_full_modelFull_estimated.mat','DCM_Subj2_fullTimeSeries_modelAlternate_estimated.mat'};

N=2;

% Specify PEB model settings (see batch editor for help on each setting)

M = struct();

M.alpha = 1;

M.beta = 16;

M.hE = 0;

M.hC = 1/16;

M.Q = 'single';

% Specify design matrix for N subjects. It should start with a constant column

M.X = ones(N,1);

% Choose field

field = {'A'};

% Estimate model

PEB = spm_dcm_peb(GCM,M,field);

save('PEB.mat','PEB');

At this stage, I tried to search over nested PEBs using (https://en.wikibooks.org/wiki/User:Peterz/sandbox)

BMA = spm_dcm_peb_bmc(PEB);

However I got an error

Error using spm_dcm_peb_bmc (line 84)

Please provide a single PEB model

I tried using comparing the full PEB to nested PEB but still got this error

BMA = spm_dcm_peb_bmc(PEB, GCM(:,1));

Error using spm_dcm_peb_bmc (line 84)

Please provide a single PEB model

Finally I used the following command and it worked

BMA = spm_dcm_peb_bmc(PEB(:,1));

Now I have 2 questions

1. Why I can only run PEB over models separately ? that means where I use PEB(:,1), I have my subjects from FullModel, and when I use PEB(:,2) I have my subjects from alternate model. How to search over nested PEBs in this case ?

2. Even with running on two subjects, I got some results, how do I know where these results are significant ?

My BMA is as follows

BMA =

struct with fields:

Snames: {2×1 cell}

Pnames: {64×1 cell}

Pind: [64×1 double]

Xnames: {'Covariate 1'}

M: [1×1 struct]

Ep: [64×1 double]

Eh: 0.7110

Ch: 0.0354

Cp: [64×64 double]

Ce: [64×64 double]

F: -1.5468e+05

Pp: [64×1 double]
My BMA.Ep is almost empty other than 3 variables, which are of the order-0.03 etc. what does that mean ? How high should the value be to consider that it exists

I know that following are the definition of the variables

DCM.Ep (averaged posterior means, $\mu$ )
DCM.Cp (averaged posterior covariance matrix, $\Lambda ^{-1}$ )
DCM.Vp (averaged posterior variance vector of each parameter - the diagonal of DCM.Cp)
DCM.Pp (averaged posterior probability of each parameter).

However I m still confused whether to identify the strength between two nodes, I would look at BMA.Ep or BMA.Cp ?

3. In the BMR window, there is a graph called posterior, It has 8 bars, and bars values are between 0.7 and 0.3, what does that mean ? It is perhaps connection strengths but can you please confirm and what strength would be considered for existence and what would be considered negligible ?

Any help will be very useful
Thank you