 |
 |
Dear Rasha, It doesn't matter whether the values are negative or positive, all that matters is the differences. A difference of more than 3 is considered strong evidence so it looks like you are consistently getting GS as the best method by far. Best, Vladimir On Wed, Oct 22, 2014 at 11:36 AM, Rasha Haider <[log in to unmask]> wrote: > Dear Vladimir, > thank you very much, your answers are always helpful for me, it worked, > but now I have another question I would appreciate if you could explain it > for me. > > After I merged the data I applied source reconstruction using the four > techniques available now in SPM12 (EBB, IID, COH, GS) to see which one is > giving me the most accurate results in order to use it for source > reconstruction for real MEG data, but unfortunately the results are almost > the same, and the problem is I got negative values for Log-evidence and > when I compared between these values for all inversion methods I didn't get > any significant difference! (please see the attached photo). > > My question is what is the meaning of the negative values of Log-evidence > and is there another way to compare the performance of inversion techniques > instead of Log-evidence? > > I hope you can help me in this matter also. > Best Regards > Rasha > > ------------------------------ > *From:* Vladimir Litvak <[log in to unmask]> > *To:* Rasha Haider <[log in to unmask]> > *Cc:* "[log in to unmask]" <[log in to unmask]> > *Sent:* Sunday, October 19, 2014 6:44 PM > > *Subject:* Re: [SPM] Concatenating several simulated Meg data sets > > Dear Rasha, > > If all your epochs are the same length you can merge them and then > re-convert your merged file forcing your data to be continuous (set 'Check > trial boundaries' to 'No'). Otherwise you'll have to use the tools for > converting arbitrary data (see the manual). > > Best, > > Vladimir > > > > On Sun, Oct 19, 2014 at 11:03 AM, Rasha Haider < > [log in to unmask]> wrote: > > Dear Gareth, > > thank you for answering my second question, but any help regarding the > main question about how to generate concatenating simulated data sets??. > I would appreciate if you or any one can help me in this matter because > this simulation is very important in my work. > > Regards > Rasha > > ------------------------------ > *From:* "Barnes, Gareth" <[log in to unmask]> > *To:* SPM (Statistical Parametric Mapping) <[log in to unmask]>; Rasha > Haider <[log in to unmask]> > *Sent:* Sunday, October 19, 2014 5:55 PM > *Subject:* Re: [SPM] Concatenating several simulated Meg data sets > > Dear Rasha > That option generates random white signals (not noise) the reason for > this is that sometimes you may wish to have a lot of signals that are not > correlated (and this is difficult to do with sinusoids). > best > Gareth > > ------------------------------ > *From:* SPM (Statistical Parametric Mapping) <[log in to unmask]> on > behalf of Rasha Haider <[log in to unmask]> > *Sent:* 19 October 2014 09:22 > *To:* [log in to unmask] > *Subject:* [SPM] Concatenating several simulated Meg data sets > > > Dear SPMer experts, > > I want to simulate MEG data set over a time window ( 0 ms - 600 ms), > but I want to divide the total time window into four sub-windows, where for > each sub-time window I simulate different sources, for this I have > simulated four MEG data sets with concatenated time windows; and I want > to Concatenate these data sets into one data set, I tried to use the merge > function but it didn't work since it requires the merged data sets to have > the same time points, and in my case the data sets have concatenated time > windows, how can I do this ? > > One more question regarding the following option in the batch editor: > > - * "Bandwidth of Gaussian orthogonal white signals in (Hz) " > > can any one explain to me what is meaning of this choice, I have used it > with the default options but I didn't get the shape of the simulated signal. > > Any help will be very appreciated. > Regards > Rasha > > > > > > > > > >