Download presentation
Presentation is loading. Please wait.
Published byOsborne Martin Modified over 9 years ago
1
Signal and Noise in fMRI fMRI Graduate Course October 15, 2003
2
What is signal? What is noise? Signal, literally defined –Amount of current in receiver coil What can we control? –Scanner properties (e.g., field strength) –Experimental task timing –Subject compliance (through training) –Head motion (to some degree) What can’t we control? –Electrical variability in scanner –Physiologic variation (e.g., heart rate) –Some head motion –Differences across subjects
3
I. Introduction to SNR
4
Signal, noise, and the General Linear Model Measured Data Amplitude (solve for) Design Model Noise Cf. Boynton et al., 1996
5
Signal-Noise-Ratio (SNR) Task-Related Variability Non-task-related Variability
6
Signal Size in fMRI 4550 50 - 45 AB C E (50-45)/45 D
9
Differences in SNR
10
Voxel 3 Voxel 2 Voxel 1 690 730 770 790 830 870 770 810 850
11
t = 16 t = 8 t = 5 A BC
12
Effects of SNR: Simulation Data Hemodynamic response –Unit amplitude –Flat prestimulus baseline Gaussian Noise –Temporally uncorrelated (white) –Noise assumed to be constant over epoch SNR varied across simulations –Max: 2.0, Min: 0.125
13
SNR = 2.0
14
SNR = 1.0
15
SNR = 0.5
16
SNR = 0.25
17
SNR = 0.125
18
SNR = 4.0SNR = 2.0 SNR = 1.0 SNR =.5
19
What are typical SNRs for fMRI data? Signal amplitude –MR units: 5-10 units (baseline: ~700) –Percent signal change: 0.5-2% Noise amplitude –MR units: 10-50 –Percent signal change: 0.5-5% SNR range –Total range: 0.1 to 4.0 –Typical: 0.2 – 0.5
20
Effects of Field Strength on SNR Turner et al., 1993
21
Theoretical Effects of Field Strength SNR = signal / noise SNR increases linearly with field strength –Signal increases with square of field strength –Noise increases linearly with field strength –A 4.0T scanner should have 2.7x SNR of 1.5T scanner T 1 and T 2 * both change with field strength –T 1 increases, reducing signal recovery –T 2 * decreases, increasing BOLD contrast
22
Measured Effects of Field Strength SNR usually increases by less than theoretical prediction –Sub-linear increases in SNR; large vessel effects may be independent of field strength Where tested, clear advantages of higher field have been demonstrated –But, physiological noise may counteract gains at high field ( > ~4.0T) Spatial extent increases with field strength Increased susceptibility artifacts
23
Excitation vs. Inhibition M1 SMA Waldvogel, et al., 2000
24
II. Properties of Noise in fMRI Can we assume Gaussian noise?
25
Types of Noise Thermal noise –Responsible for variation in background –Eddy currents, scanner heating Power fluctuations –Typically caused by scanner problems Variation in subject cognition –Timing of processes Head motion effects Physiological changes Differences across brain regions –Functional differences –Large vessel effects Artifact-induced problems
26
Why is noise assumed to be Gaussian? Central limit theorem
27
Is noise constant through time?
29
Is fMRI noise Gaussian (over time)?
30
Is Signal Gaussian (over voxels)?
31
Variability
32
Variability in Subject Behavior: Issues Cognitive processes are not static –May take time to engage –Often variable across trials –Subjects’ attention/arousal wax and wane Subjects adopt different strategies –Feedback- or sequence-based –Problem-solving methods Subjects engage in non-task cognition –Non-task periods do not have the absence of thinking What can we do about these problems?
33
Response Time Variability AB
34
Intersubject Variability A & B: Responses across subjects for 2 sessions C & D: Responses within single subjects across days E & F: Responses within single subjects within a session - Aguirre et al., 1998 B A C D E F
35
Variability Across Subjects D’Esposito et al., 1999
36
Young Adults
37
Elderly Adults
40
Effects of Intersubject Variability
41
Parrish et al., 2000
42
Implications of Inter-Subject Variability Use of individual subject’s hemodynamic responses –Corrects for differences in latency/shape Suggests iterative HDR analysis –Initial analyses use canonical HDR –Functional ROIs drawn, interrogated for new HDR –Repeat until convergence Requires appropriate statistical measures –Random effects analyses –Use statistical tests across subjects as dependent measure (rather than averaged data)
43
Spatial Variability? AB McGonigle et al., 2000
44
Standard Deviation Image
45
Spatial Distribution of Noise A: Anatomical Image B: Noise image C: Physiological noise D: Motion-related noise E: Phantom (all noise) F: Phantom (Physiological) - Kruger & Glover (2001)
46
Low Frequency Noise
47
High Frequency Noise
48
III. Methods for Improving SNR
49
Fundamental Rule of SNR For Gaussian noise, experimental power increases with the square root of the number of observations
51
Trial Averaging Static signal, variable noise –Assumes that the MR data recorded on each trial are composed of a signal + (random) noise Effects of averaging –Signal is present on every trial, so it remains constant through averaging –Noise randomly varies across trials, so it decreases with averaging –Thus, SNR increases with averaging
52
Example of Trial Averaging Average of 16 trials with SNR = 0.6
53
1 2 3 4 5 6
55
Increasing Power increases Spatial Extent Subject 1Subject 2 Trials Averaged 4 16 36 64 100 144 500 ms 16-20 s 500 ms …
56
AB
57
Number of Trials Averaged Number of Significant Voxels Subject 1 Subject 2 V N = V max [1 - e (-0.016 * N) ] Effects of Signal-Noise Ratio on extent of activation: Empirical Data
58
Active Voxel Simulation Signal + Noise (SNR = 1.0) Noise 1000 Voxels, 100 Active Signal waveform taken from observed data. Signal amplitude distribution: Gamma (observed). Assumed Gaussian white noise.
59
Effects of Signal-Noise Ratio on extent of activation: Simulation Data SNR = 0.10 SNR = 0.15 SNR = 0.25 SNR = 1.00 SNR = 0.52 (Young) SNR = 0.35 (Old) Number of Trials Averaged Number of Activated Voxels
60
Explicit and Implicit Signal Averaging r =.42; t(129) = 5.3; p <.0001 r =.82; t(10) = 4.3; p <.001 A B
61
Caveats Signal averaging is based on assumptions –Data = signal + temporally invariant noise –Noise is uncorrelated over time If assumptions are violated, then averaging ignores potentially valuable information –Amount of noise varies over time –Some noise is temporally correlated (physiology) Nevertheless, averaging provides robust, reliable method for determining brain activity
62
Accurate Temporal Sampling
63
Visual HDR sampled with a 1-sec TR
64
Visual HDR sampled with a 2-sec TR
65
Visual HDR sampled with a 3-sec TR
66
Comparison of Visual HDR sampled with 1,2, and 3-sec TR
67
Visual HDRs with 10% diff sampled with a 1-sec TR
68
Visual HDR with 10% diff sampled with a 3-sec TR
69
Accurate Spatial Sampling
70
Partial Volume Effects
75
Where are partial volume effects most problematic? Ventricles Grey / white boundary Blood vessels
76
Activation Profiles White Matter Gray / White Ventricle
77
Temporal Filtering
78
Filtering Approaches Identify unwanted frequency variation –Drift (low-frequency) –Physiology (high-frequency) –Task overlap (high-frequency) Reduce power around those frequencies through application of filters Potential problem: removal of frequencies composing response of interest
79
Power Spectra AB
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.