Title: The geometry of random images in astrophysics and brain mapping
1The geometry of random images in astrophysics and
brain mapping
- Keith Worsley, McGill University, Montreal
- Jin Cao, Lucent Technologies
- David Siegmund, Stanford
- Khalil Shafie, Shahid Beheshti University, Tehran
- Jonathan Taylor, McGill and Stanford
- Louis Collins, Jason Lerch, David MacDonald,
Tomas Paus, Alex Zijdenbos, Alan Evans, - Montreal Neurological Institute
- www.math.mcgill.ca/keith
2(No Transcript)
3fMRI data 120 scans, 3 scans each of hot, rest,
warm, rest, hot, rest,
X (hot warm effect) / S.d. N(0,1) if no
effect
4One cycle, averaged over 10 cycles
892
fMRI data
890
Hot
Simple idea fit b1 sin b2 cos test b10,
b20, using X Z12 Z22 where Zi bi/sd(bi)
888
886
delay and spreading 5(?) secs
884
882
Warm
880
878
Rest
Rest
876
874
872
0
5
10
15
20
25
30
35
40
45
Time, seconds
5(No Transcript)
6PCA_IMAGE PCA of time ? space
1 exclude first frames
2 drift
3 long-range correlation or anatomical effect
remove by converting to of brain
4 signal?
7FMRILM fits a linear model for fMRI time series
with AR(p) errors
- Linear model
- ?
? - Yt (stimulust HRF) b driftt c errort
- AR(p) errors
- ? ?
? - errort a1 errort-1 ap errort-p s WNt
unknown parameters
8(No Transcript)
9FMRIDESIGN example pain perception
10(No Transcript)
11FMRILM first step estimate the autocorrelation
?
- AR(1) model errort a1 errort-1 s WNt
- Fit the linear model using least squares
- errort Yt fitted Yt
- â1 Correlation ( errort , errort-1)
- Estimating errorts changes their correlation
structure slightly, so â1 is slightly biased - Raw autocorrelation Smoothed 15mm Bias
corrected â1 -
-0.05 0
12FMRILM second step refit the linear model
Pre-whiten Yt Yt â1 Yt-1, then fit using
least squares
13Higher order AR model? Try AR(3)
a
a
a
1
2
3
0.3
0.2
AR(1) seems to be adequate
0.1
0
has little effect on the T statistics
-0.1
AR(1)
AR(2)
AR(3)
5
0
-5
14Milky way
15(No Transcript)
16Gaussianized T field?
EC
Gaussian threshold
17EC for WMAP, unsmoothed, FWHM 1.6o
2500
Observed
2000
Expected
10
1500
0
1000
500
-10
-8
-7
-6
-5
-4
0
EC
20
-500
-1000
10
-1500
0
4
5
6
7
8
-2000
-2500
-5
-4
-3
-2
-1
0
1
2
3
4
5
Gaussian threshold
18Observed Expected EC, with Poisson sd in tails
EC
Gaussian threshold
19EC for WMAP smoothed to FWHMo (expected)
no smoothing
scale space
20Simulation (expected)
no smoothing
scale space
21Scale space WMAP data
2000
Observed
Expected
o
o
1.6
20.5
1500
1000
10
0
500
EC
-10
-4.5
-4
-3.5
0
20
10
-500
0
3.5
4
4.5
5
-1000
-5
-4
-3
-2
-1
0
1
2
3
4
5
Gaussian threshold
22Scale space smooth X(t) with a range of filter
widths, s continuous wavelet transform adds an
extra dimension to the random field X(t, s)
Scale space, no signal
34
8
22.7
6
4
15.2
2
10.2
0
-2
6.8
-60
-40
-20
0
20
40
60
S FWHM (mm, on log scale)
One 15mm signal
34
8
22.7
6
4
15.2
2
10.2
0
-2
6.8
-60
-40
-20
0
20
40
60
t (mm)
15mm signal best detected with a 15mm smoothing
filter
23Matched Filter Theorem ( Gauss-Markov Theorem)
to best detect a signal white noise, filter
should match signal
10mm and 23mm signals
34
8
22.7
6
4
15.2
2
10.2
0
-2
6.8
-60
-40
-20
0
20
40
60
S FWHM (mm, on log scale)
Two 10mm signals 20mm apart
34
8
22.7
6
4
15.2
2
10.2
0
-2
6.8
-60
-40
-20
0
20
40
60
t (mm)
But if the signals are too close together they
are detected as a single signal half way between
them
24Scale space can even separate two signals at the
same location!
8mm and 150mm signals at the same location
10
5
0
-60
-40
-20
0
20
40
60
170
113.7
20
76
50.8
15
S FWHM (mm, on log scale)
34
10
22.7
15.2
5
10.2
6.8
-60
-40
-20
0
20
40
60
t (mm)
25FWHM 6.8mm
26FWHM 9mm
27FWHM 11mm
28FWHM 15mm
29FWHM 20mm
30FWHM 26mm
31FWHM 34mm
32FWHM
33FWHM
34FWHM
35FWHM
36FWHM
37FWHM
38(No Transcript)
39FWHM
40FWHM
41fMRI data 120 scans, 3 scans each of hot, rest,
warm, rest, hot, rest,
X (hot warm effect) / S.d. N(0,1) if no
effect
42One cycle, averaged over 10 cycles
892
fMRI data
890
Hot
Simple idea fit b1 sin b2 cos test b10,
b20, using X Z12 Z22 where Zi bi/sd(bi)
888
886
delay and spreading 5(?) secs
884
882
Warm
880
878
Rest
Rest
876
874
872
0
5
10
15
20
25
30
35
40
45
Time, seconds