fMRI data and SPM2

1
fMRI data and SPM2

• introduction to the SPM software

2
2
BOLD signal and fMRI time-series
Time
Time
Intensity
single voxel time series
3
image data
parameter estimates
designmatrix
kernel

• General Linear Model
• model fitting
• statistic image

realignment motioncorrection
random field theory
smoothing
normalisation
StatisticalParametric Map
anatomicalreference
corrected p-values
4
Movement Correction Why?

• Sensitivity Large error variance may prevent us
  from finding activations.
• Specificity Task correlated motion may pose as
  activations.

Large Activation
Intensity in voxel
Scan
5
Possibly correct for distortions
Distorted image
Corrected image
Correction
6
Spatial Normalisation
Non-linear registration
Affine registration
7
Regression example
Ys ? ? f(s) ?s
f(s) 0 or 1
?s N(0,?2)

?
?

error

• t-statistic for H0 ??gt 0

• account for temporal autocorrelation

• correlationtest H0 ? 0 eqivalent totest H0
  ? 0

box-car waveform
voxel time series
8
revisited

??
?


error
?s

??
?

f1(ts)
1
Ys

9
Inference contrasts SPMt
contrast linear combination of parameters c ?
c 1 0 0 0 0 0 0 0
test H0 c ? 0
e.g. box-car amplitude gt 0 ?
contrast ofestimatedparameters
T
varianceestimate
SPMt
10
Restricted Maximum Likelihood
observed
Q1
ReML
estimated
Q2
11
Using an uncorrected p-value of 0.1 will lead
us to conclude on average that 10 of voxels are
active when they are not.
This is clearly undesirable. To correct for this
we can define a null hypothesis for images of
statistics.
12
Use of uncorrected p-value, a0.1
Use of corrected p-value, a0.1
FWE