Hi Payal,
yes, time *always* starts at zero, regardless whether you count in
units of scans or seconds. Names of your image files or frames of a 4D
dataset do not matter, as long as you enter them in chronological
order.
Best,
Volkmar
Am Dienstag, den 11.12.2018, 10:04 -0500 schrieb Payal Arya:
> Hi Volkmar
>
> Thank you so much for your reply. I will make the necessary changes.
> I just had one more question. I saw that in some other email thread ,
> you mentioned even if i specify design in the first level in terms of
> scans, SPM counts first scan at t=0. Does this mean that while
> putting in scan onset in terms of scan numbers- first scan number
> should be 0?
>
> I have a block design and i do two scans per block. I use scan
> numbers as onset in design specification. Block design goes as
> silence (2 scans ) ,Task1 (two scans),silence(2scans),task2(2scans).
> So onset for silence -[1 5]
> Task 1- [3 11]
> Task2-[7 15]
>
> Also does it matter how the file numbers are named if the scan
> numbers have to start from 0 ?
> My scan file are named as vol0001.nii----vol0450.nii.
>
>
> Thank you
>
> Best
> Payal
>
> On Mon, Dec 10, 2018 at 5:09 AM Volkmar Glauche <volkmar.glauche@unik
> linik-freiburg.de> wrote:
> > Hi Payal,
> >
> > if you are applying smoothing before upscaling, 0.5mm FWHM might be
> > sufficient. Since your voxel size is larger than that in z
> > direction,
> > SPM may still come up with a higher smoothness estimate in z
> > compared
> > to x and y.
> > As a side note: I would probably do the upscaling as the very first
> > preprocessing step and use 5mm smoothing. Since this only involves
> > changing the voxel-to-world mapping matrix, it does not require any
> > resampling of the image data. By treating the upscaled data as your
> > raw
> > data, you can be sure not to forget upscaling later on if you need
> > to
> > adjust your preprocessing pipeline.
> >
> > Best,
> > Volkmar
> >
> > Am Donnerstag, den 06.12.2018, 13:07 -0500 schrieb Payal Arya:
> > > Dear Volkmar
> > >
> > > I am sorry I should have mentioned the units in my previous
> > mail.
> > >
> > > The matrix size of atlas is 64x64x16, fov- 204.8 mm
> > x204.mm8x204.8mm,
> > > voxels- 3.2mmx3.2mmx12.8mm
> > >
> > > The matrix size of functional data is 64x64x15 , fov- 25mm x
> > 25mmx
> > > 11.25mm and voxel spacing is-0.39 mmx0.39mmx0.75mm.
> > > I applied the a smoothing kernel of 0.5 mm in each direction to
> > the
> > > functional data before registering it with atlas.
> > >
> > > After preprocessing and smoothing I upscaled the functional data
> > by a
> > > factor of 10 (voxels and fov) and then registered it to the
> > atlas.
> > >
> > > We used Spin echo pulse sequence with a RARE factor of 8 to
> > obtain
> > > functional data on a 9.4 T bruker system. What I do know is
> > that
> > > with SE pulse sequences, the extent os bold response or the
> > signal is
> > > low as well as its not noisy like EPI.
> > > So I am confused as to what would be a good smoothing kernel so
> > that
> > > i do not loose signal.
> > >
> > >
> > > Thank you for your help.
> > >
> > > Best regards
> > >
> > > Payal
> > >
> > >
> > > On Thu, Dec 6, 2018 at 4:44 AM Volkmar Glauche <volkmar.glauche@u
> > nikl
> > > inik-freiburg.de> wrote:
> > > > Dear Payal,
> > > > I do not have sufficient knowledge about the spatial
> > > > characteristics of songbird brains, their BOLD effect
> > > > characteristics and the spatial resolution of your data (you
> > > > mentioned numbers, but no units). In humans, the recommendation
> > is
> > > > to use a smoothing kernel with FWHM ~2-3 times voxel size.
> > Reasons
> > > > for this recommendation include good match of spatial extent of
> > > > BOLD effect, spatial extent of activated brain structures and
> > > > accounting for inter-subject variability in localisation of
> > brain
> > > > function.
> > > > Hope this helps
> > > > Volkmar
|