Dear Peichi
| I am trying to use fMRI to see the V1 response to different
|frequency of flashing checkboard using block design as follows:
| blank(15 scans) => 1Hz(15 scans) =>blank(15 scans)=> 4Hz(15 scans)=>
|blank(15 scans) =>7Hz(15 scans)
| Can anybody tell me how to use SPM99b to perform regression and see
|the resonse of V1 to different flashing frequencies
| Thanks
I believe you can use the 'parametric specification' dialogue when
specifying the design matrix to do this.
Specify your design matrix intially as if this were a single condition
experiment (active vs rest - see previous postings about how to do
this).Then when you reach the 'parametric specification' dialogue, press
'other'. This section allows you to specify a parameter pertaining to any
(or all) of the trial(epoch) types. A design matrix will be constructed
that reflects the interaction between this parameter and the trial (epoch)
types you specify. In your case the parameter of interest is reversal
frequency.
For the simplest case (testing for a linear relationship), you would pick
'linear' for the expansion, and specify three values for your parameter
that reflect the parameter (reversal frequency) of this trial type i.e. [1
4 7]. This will construct a three column design matrix with the session
effect, the constant (intercept) term and the (mean corrected) linear
relationship. Testing for more complex non-linear relationships
(such as might be expected in V1 above 8Hz) is also possible by selecting
a polynomial or exponential expansion. For more detail on this see
Christian Buchel's paper (Neuroimage 1996 Aug;4(1):60-6)
best wishes,
Geraint
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|