 |
 |
Dear Tibor, The simulation is done by specifying the change in T2*. This is the physical quantity that changes. In order to model how this relates to the stimulation requires something like a double-gamma HRF convolution, a more sophisticated balloon model, or something else. As there is fully correct, understood and validated way to predict the BOLD response from the stimulation (this is an active field of research) the simulation does not impose a model. It requires the user to select a model appropriate for their own purposes and then simply specify the result, in terms of T2* changes. I hope this is clear. If you are happy with a standard HRF model, then convolving with a double-gamma (or even a single gamma) and then converting signal change into T2* is what is required. As for the time - I'm not surprised by the 5.9 days. With your resolution and number of time points it is an *intensive* calculation process. Also this is only an estimate, so it could be more than this too. It is based on timing tests done with our server, so it depends on the relative performance of those machines. In order to utilize other cores in your cluster you must install and use SGE. The FMRIB web pages and email list has details on setting this up. I still strongly recommend that you do a small test (e.g. one slice) before proceeding to anything larger. All the best, Mark On 19 Apr 2010, at 15:10, Auer, Tibor MD. Ph.D. wrote: > Dear Mark, > > Thank you for your description. I have seen this > activation3Dtimecourse-file. But it seems to be just a simple > BOLD-signal-change (double-gamma, I guess), which does not describe a > timecourse of a block-design. > How should I alter this file to get volumes "acquired" for a simple > block-design (e.g.: 18s passive, 12s active, 8 cycles; 129 volumes > assuming > TR=2s)? > > For the calculation time: it really says 5.9 days. Our server has > eight > cores. How can I use all of them? > > Auer, Tibor M.D. Ph.D. > Biomedizinische NMR Forschungs GmbH > am Max-Planck Institut für Biophysikalische Chemie > Am Fassberg 11 > 37077 Göttingen > Germany > Phone/Work: +49-(0)551-201-1725 > Phone/Home: +49-(0)551-250-4576 > Mobile: +49-(0)176-8012-7921 > Mail: [log in to unmask] > > -----Original Message----- > From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On > Behalf > Of Mark Jenkinson > Sent: Friday, April 16, 2010 4:42 PM > To: [log in to unmask] > Subject: Re: [FSL] FMRI-Simluation > > Dear Tibor, > > Have you been using the GUI? > There is a fairly detailed documentation page at: > http://www.fmrib.ox.ac.uk/fsl/possum/ > which takes you through the different inputs > required. The GUI should also give you an > estimate of the running time at the bottom > right. What did it say? Doing a full fMRI > simulation is *very* intensive and I suspect > that with this kind of resolution you are looking > at a very long time on a single machine (much > more than a week!) but on a sizable cluster it > should be feasible. > > I recommend trying a small scale test to begin > with (e.g. one slice) to check that you have > got things set up correctly. > > We do not currently have a module devoted to > POSSUM in the FSL course and so it does not > have the same information there (i.e. Tutorial/Practical) > like many of the other tools have. > > In the documentation at: > http://www.fmrib.ox.ac.uk/fsl/possum/input.html#activation > it describes how the activation is modeled via T2* values > and points you to example files. If you look at the > activation3Dtimecourse > file you will see that it is a simple text file with two columns. > The first is time (in seconds) the second is the T2* value. > Here the T2* value has been fit experimentally and is quite > approximate, but theoretical values can be used too. The > most important equation is S/S0 = exp(-TE/T2*) which tells > you how the signal (S), as a fraction of the baseline signal, S0, > varies as a function of TE (specified in the pulse sequence) > and T2* (which is variable and specified in the appropriate > file). To get a certain change in intensity (S/S0) you can > work out what T2* change is required for a given TE. > The other input is a 3D image, specifying the weighting of > the T2* change at each voxel. This is typically a map that > specifies the activation location, and can be a binary image > or a more realistic case where the "strength" of the activation > drops off at the boundaries. Look at the example to see what > it looks like, although you'll need to reset the display range > in FSLView to -0.01 to 0.01 (or similar) as the values are > quite small (which modulates the T2* change, giving an > overall T2* change of approximately 6ms in the active regions). > Note that the T2* specified here is always added to the > base T2* value of that tissue type (which is 51ms for grey > matter at 3T and 83ms at 1.5T - see the $FSLDIR/data/possum/MRpar_3T > and $FSLDIR/data/possum/MRpar_1.5T files which are > also described in the documentation). > > Note also that any timing can be specified and that it > linearly interpolates between T2* values at the specified > times. > > I hope this is helpful. > All the best, > Mark > > > > On 16 Apr 2010, at 15:17, Auer, Tibor MD. Ph.D. wrote: > >> Dear Experts, >> >> I am trying to find a software to perform fMRI simulations. I set up >> 100 volumes each containing 18 slices with resolution of 2mmx2mm. >> In FSL-package, there is POSSUM available. However, there is no >> User Guide/Tutorial/Practical guide for POSSUM similar to that given >> for other FSL tools. I also tried to look for MIDAS, but the given >> website (http://www.midas-online.org/) is some crap about printers >> (?). >> >> I tried also to figure out POSSUM myself, so I gave simply the >> parameters and the supplied activation maps. However, I do not know >> how to set up the paradigm (control and active phases). It did not >> give me any failure message (I have checked the logfile), but it >> runs for almost a week. >> How can I define custom activation pattern (shape, size and >> amplitude)? How can I define paradigm? >> >> Or is there any useful tool for fMRI-data simulation? >> >> Thanks, >> >> Auer, Tibor M.D. Ph.D. >> Biomedizinische NMR Forschungs GmbH >> am Max-Planck Institut für Biophysikalische Chemie >> Am Fassberg 11 >> 37077 Göttingen >> Germany >> Phone/Work: +49-(0)551-201-1725 >> Phone/Home: +49-(0)551-250-4576 >> Mobile: +49-(0)176-8012-7921 >> Mail: [log in to unmask] >> >