Dear Juha,
The 4D activation volume must contain the change in
T2* values at each voxel, at each point in time. That is,
the T2* value minus the static value specified in the
MR parameter file for that tissue. So a value of 0
simply means that the "normal" value of T2* is used.
Any positive or negative values are added to that value.
Please note that the memory and computational requirements
for the 4D activation volume are really quite high, so
I strongly recommend that you try out your command
on a small test case first to verify that it is working and
then progressively scale up from that.
All the best,
Mark
On 21 Jun 2010, at 12:05, Juha Pajula wrote:
> Thanks for your advices
>
> I still need to ask again, while the 3D activation is not enough for
> our
> study. Possum does not have examples for 4D activation volumes and
> timecourse.
>
> If I understood right the timecourse in 4D activation includes only
> the time
> points of the activation volumes (according of possum command).
>
> The real question is then the format of the 4D activation volume.
> Does the
> intensities in the activation volumes need to be scaled in some
> certain
> level, for example [-1 1]? Are every volume in the 4D activation
> independent or do they depend somehow from each others (for example,
> first
> volume defines the activation intensities and the next just includes
> the
> changes from the first)?
>
> We need multiple different activation patterns in the same
> simulation and
> so the 4D activation is our only choice. I have created CLI tool to
> use
> possum easier with 4D activation without GUI, so the limitations of
> the
> GUI are not a problem.
>
> - Juha
>
>> Dear Juha,
>>
>> What you need to specify in possum is the change in T2* values.
>> This relates to the intensity of the signal (assuming nothing
>> is moving) via the equation:
>> S = S0 * exp(-TE/T2*)
>>
>> Hence you need to specify time-varying T2* values.
>> Look at the files $FSLDIR/data/possum/activation3D.nii.gz
>> and $FSLDIR/data/possum/activation3Dtimecourse to see
>> the format of the spatial map and timecourse. The first
>> column of the second file specifies values of time (in seconds)
>> and the second specifies the change in T2* values (from
>> the fixed, static value specified in the MR parameter file).
>> The units for both of these things is seconds, not milliseconds.
>>
>> Note that the HRF is not added for you - it is necessary for
>> you to specify a timecourse that already takes this into account.
>>
>> The specification of procnum works fine at our end, so I'm
>> not sure why it is not working for you. Feel free to modify it
>> if that works for you.
>>
>> As for the memory and time requirements for possum - they
>> can be very high if you specify very long time series and/or
>> high resolutions. I strongly recommend running a small test
>> case first and then seeing how it scales up.
>>
>> I hope this helps.
>> All the best,
>> Mark
>>
>>
>>
>> On 15 Jun 2010, at 15:43, Juha Pajula wrote:
>>
>>> Hi!
>>>
>>> I am creating 4D activation sequences to POSSUM and now I need some
>>> advices.
>>>
>>> I have fMRI data from previous studies and now we need to create
>>> activations for POSSUM from the data. I found some instructions from
>>> the POSSUM homepage, but could you give bit more detailed
>>> instructions here; especially in the case of 4D activation.
>>>
>>> How the data should be scaled? Is the data impressed separately in
>>> every volume or do the following volumes after the first one having
>>> just the changes from the first activation?
>>>
>>> Do I need to consider the effect of haemodynamic response function?
>>>
>>> What does the following error truly means:
>>> ------
>>> Creating activation4D volume and timecourse vector...
>>> 4D activation mode
>>> Testing if the values in the activation volumes are smaller than the
>>> T2* values
>>> vol= 0
>>> WARNING: Perturbation in the T2* values bigger than the T2* values.
>>> ------
>>>
>>> I sent one email at last week about the memory consumption of the
>>> POSSUM simulator processes in the SGE. Can you give any advices to
>>> that now?
>>>
>>>
>>> I think I found a bug in possumX in fsl version 4.1.4 from the line
>>> 240:
>>> procnum=0
>>>
>>> I guess it should be
>>> procnum=1
>>>
>>> With the 0 the script creates one extra process to the SGE, which is
>>> disturbing if the maximum amount of processes per user is limited in
>>> the system. Could you check that it does not have any other effects
>>> to the simulation?
>>>
>>>
>>> --
>>> Juha Pajula,
>>> Research Assistant,
>>> Methods and Models for Biological Signals and Images group of Signal
>>> Processing department in Tampere University of Technology
>>>
>>
>
|