ruff spm efmri fil 2011 - fil.ion.ucl.ac.uk · spm{f} timing issues: slice timing “slice-timing...
TRANSCRIPT
![Page 1: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/1.jpg)
Event-related fMRI
Christian Ruff
Laboratory for Social and Neural Systems ResearchDepartment of Economics
University of Zurich
Institute of NeurologyUniversity College London
With thanks to the FIL methods group, in particular Rik Henson
![Page 2: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/2.jpg)
Realignment Smoothing
Normalisation
General linear model
Statistical parametric map (SPM)Image time-series
Parameter estimates
Design matrix
Template
Kernel
Gaussian field theory
p <0.05
Statisticalinference
![Page 3: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/3.jpg)
Overview
1. Block/epoch vs. event-related fMRI
2. (Dis)Advantages of efMRI
3. GLM: Convolution
4. BOLD impulse response
5. Temporal Basis Functions
6. Timing Issues
7. Design Optimisation – “Efficiency”
![Page 4: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/4.jpg)
U1 P1 U3U2 P2
Data
Model
P = PleasantU = Unpleasant
Block/epoch designs examine responses to series of similar stimuli
U1 U2 U3 P1 P2 P3
Event-related designs account for response to each single stimulus
~4s
Block/epoch designs vs event-related designs
![Page 5: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/5.jpg)
Advantages of event-related fMRI
1. Randomised trial order
![Page 6: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/6.jpg)
Blocked designs may trigger expectations and cognitive sets
…Pleasant (P)Unpleasant (U)
efMRI: Randomised trial order
Intermixed designs can minimise this by stimulus randomisation
… … ………
Unpleasant (U) Unpleasant (U) Unpleasant (U)Pleasant (P) Pleasant (P)
![Page 7: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/7.jpg)
1. Randomised trials order
2. Post-hoc subjective classification of trials
Advantages of event-related fMRI
![Page 8: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/8.jpg)
Gonsalves & Paller (2000) Nature Neuroscience
Items with wrong memory of picture („hat“) were associated with
more occipital activity at encoding than items with correct rejection („brain“)
„was shown as picture“
„was not shown as picture“
Participant response:
efMRI: Post-hoc classification of trials
![Page 9: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/9.jpg)
1. Randomised trials order
2. Post-hoc subjective classification of trials
3. Some events can only be indicated by participant
Advantages of event-related fMRI
![Page 10: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/10.jpg)
efMRI: Online event definition
![Page 11: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/11.jpg)
1. Randomised trials order
2. Post-hoc subjective classification of trials
3. Some events can only be indicated by participant
4. Some events cannot be blocked due to stimulus context
Advantages of event-related fMRI
![Page 12: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/12.jpg)
…time
Oddball
efMRI: Stimulus context
![Page 13: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/13.jpg)
1. Randomised trials order
2. Post-hoc subjective classification of trials
3. Some events can only be indicated by participant
4. Some events cannot be blocked due to stimulus context
5. More accurate model even for epoch/block designs?
Advantages of event-related fMRI
![Page 14: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/14.jpg)
“Event” model may capture state-item interactions (with longer SOAs)
“Epoch” model assumes constant neural processes throughout block
DataModel
P = PleasantU = Unpleasant
U1 U2 U3 P1 P2 P3
U1 U2 U3 P1 P2 P3
“Event” model of block design
![Page 15: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/15.jpg)
Convolved with HRF
=>
Series of eventsDelta
functions
“Classic” Boxcar function
Sustained epochDesigns can be blocked or intermixed, BUT models for blocked designs can be epoch- or event-related
Epochs are periods of sustained stimulation (e.g, box-car functions)Events are impulses (delta-functions)
Near-identical regressors can be created by 1) sustained epochs, 2) rapid series of events (SOAs<~3s)
In SPM8, all conditions are specified in terms of their 1) onsets and 2) durations … epochs: variable or constant duration … events: zero duration
Modeling block designs: Epochs vs events
![Page 16: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/16.jpg)
β=3 β=5
β=9β=11
Rate = 1/4s Rate = 1/2s
Modeling block designs: Epochs vs events
• Blocks of trials can be modeled as boxcars or runs of events
• BUT: interpretation of the parameter estimates may differ
• Consider an experiment presenting words at different rates in different blocks:
‣ An “epoch” model will estimate parameter that increases with rate, because the parameter reflects response per block
‣ An “event” model may estimate parameter that decreases with rate, because the parameter reflects response per word
![Page 17: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/17.jpg)
Disadvatages of intermixed designs
1. Less efficient for detecting effects than blocked designs (see later…)
2. Some psychological processes have to/may be better blocked (e.g., if difficult to switch between states, or to reduce surprise effects)
![Page 18: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/18.jpg)
Overview
1. Block/epoch vs. event-related fMRI
2. (Dis)Advantages of efMRI
3. GLM: Convolution
4. BOLD impulse response
5. Temporal Basis Functions
6. Timing Issues
7. Design Optimisation – “Efficiency”
![Page 19: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/19.jpg)
BriefStimulus
Undershoot
InitialUndershoot
Peak
BOLD impulse response
• Function of blood oxygenation, flow, volume
• Peak (max. oxygenation) 4-6s poststimulus; baseline after 20-30s
• Initial undershoot can be observed
• Similar across V1, A1, S1…
• … but possible differences across: - other regions - individuals
![Page 20: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/20.jpg)
BriefStimulus
Undershoot
InitialUndershoot
Peak
BOLD impulse response
• Early event-related fMRI studies used a long Stimulus Onset Asynchrony (SOA) to allow BOLD response to return to baseline
• However, overlap between successive responses at short SOAs can be accommodated if the BOLD response is explicitly modeled, particularly if responses are assumed to superpose linearly
• Short SOAs are more sensitive; see later
![Page 21: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/21.jpg)
GLM for a single voxel:
y(t) = u(t) ⊗ h(τ) + ε(t)
u(t) = neural causes (stimulus train)
u(t) = ∑ δ (t - nT)
h(τ) = hemodynamic (BOLD) response
h(τ) = ∑ ßi fi (τ)
fi(τ) = temporal basis functions
y(t) = ∑ ∑ ßi fi (t - nT) + ε(t)
y = X ß + ε
Design Matrix
convolution
T 2T 3T ...
u(t) h(τ)=∑ ßi fi (τ)
sampled each scan
General Linear (Convolution) Model
![Page 22: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/22.jpg)
Stimulusevery 20s
SPM{F}
0 time {secs} 30
Sampled every TR = 1.7s Design matrix, X
[x(t)⊗ƒ1(τ) | x(t)⊗ƒ2(τ) |...]…
Gamma functions ƒi(τ) of peristimulus time τ(Orthogonalised)
General Linear Model in SPM
![Page 23: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/23.jpg)
Overview
1. Block/epoch vs. event-related fMRI
2. (Dis)Advantages of efMRI
3. GLM: Convolution
4. BOLD impulse response
5. Temporal Basis Functions
6. Timing Issues
7. Design Optimisation – “Efficiency”
![Page 24: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/24.jpg)
Temporal basis functionsTemporal basis functions
![Page 25: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/25.jpg)
• Fourier Set- Windowed sines & cosines- Any shape (up to frequency limit)- Inference via F-test
• Finite Impulse Response- Mini “timebins” (selective averaging)- Any shape (up to bin-width)- Inference via F-test
Temporal basis functions
![Page 26: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/26.jpg)
• Fourier Set / FIR- Any shape (up to frequency limit / bin width)- Inference via F-test
• Gamma Functions- Bounded, asymmetrical (like BOLD)- Set of different lags- Inference via F-test
• “Informed” Basis Set- Best guess of canonical BOLD response - Variability captured by Taylor expansion - “Magnitude” inferences via t-test…?
Temporal basis functions
![Page 27: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/27.jpg)
Canonical
Informed basis set
• Canonical HRF (2 gamma functions)Canonical
![Page 28: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/28.jpg)
Canonical
Informed basis set
• Canonical HRF (2 gamma functions)
plus Multivariate Taylor expansion in:- time (Temporal Derivative)
CanonicalTemporal
![Page 29: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/29.jpg)
Canonical
Informed basis set
CanonicalTemporal
• Canonical HRF (2 gamma functions)
plus Multivariate Taylor expansion in:- time (Temporal Derivative)
![Page 30: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/30.jpg)
Canonical
Informed basis set
• Canonical HRF (2 gamma functions)
plus Multivariate Taylor expansion in:- time (Temporal Derivative)- width (Dispersion Derivative)
CanonicalTemporalDispersion
![Page 31: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/31.jpg)
Canonical
Informed basis set
• Canonical HRF (2 gamma functions)
plus Multivariate Taylor expansion in:- time (Temporal Derivative)- width (Dispersion Derivative)
CanonicalTemporalDispersion
![Page 32: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/32.jpg)
Canonical
Informed basis set
• “Latency” inferences via tests on ratio of derivative : canonical parameters
• “Magnitude” inferences via t-test on canonical parameters (providing canonical is a reasonable fit)
CanonicalTemporalDispersion
• Canonical HRF (2 gamma functions)
plus Multivariate Taylor expansion in:- time (Temporal Derivative)- width (Dispersion Derivative)
![Page 33: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/33.jpg)
+ FIR+ Dispersion+ TemporalCanonical
… canonical + temporal + dispersion derivatives appear sufficient to capture most activity… may not be true for more complex trials (e.g. stimulus-prolonged delay (>~2 s)-response)… but then such trials better modelled with separate neural components (i.e., activity no longer delta function) + constrained HRF
In this example (rapid motor response to faces, Henson et al, 2001)…
Which temporal basis set?
![Page 34: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/34.jpg)
Overview
1. Block/epoch vs. event-related fMRI
2. (Dis)Advantages of efMRI
3. GLM: Convolution
4. BOLD impulse response
5. Temporal Basis Functions
6. Timing Issues
7. Design Optimisation – “Efficiency”
![Page 35: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/35.jpg)
Timing issues: Sampling
Scans TR=4s• Typical TR for 60 slice EPI at 3 mm
spacing is ~ 3-4s
![Page 36: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/36.jpg)
Timing issues: Sampling
Scans TR=4s• Typical TR for 60 slice EPI at 3 mm
spacing is ~ 3-4s
• Sampling at [0,4,8,12…] post- stimulus may miss peak signal
Stimulus (synchronous) SOA=8s
Sampling rate=4s
![Page 37: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/37.jpg)
Timing issues: Sampling
Scans TR=4s• Typical TR for 60 slice EPI at 3 mm
spacing is ~ 3-4s
• Sampling at [0,4,8,12…] post- stimulus may miss peak signal
• Higher effective sampling by: 1. Asynchrony; e.g., SOA=1.5TR 2. Random Jitter; e.g., SOA=(2±0.5)TR
• Better response characterisation
Stimulus (random jitter)
Sampling rate=2s
![Page 38: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/38.jpg)
x2 x3
T=16, TR=2s
Scan0 1
o
T0=9 oT0=16
T1 = 0 s
T16 = 2 s
Timing issues: Slice Timing
![Page 39: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/39.jpg)
Bottom SliceTop Slice
SPM{t} SPM{t}
TR=3s
Interpolated
SPM{t}
Derivative
SPM{F}
Timing issues: Slice Timing
“Slice-timing Problem”:‣ Slices acquired at different times, yet
model is the same for all slices
‣ different results (using canonical HRF) for different reference slices
‣ (slightly less problematic if middle slice is selected as reference, and with short TRs)
Solutions:1. Temporal interpolation of data
… but less good for longer TRs
2. More general basis set (e.g., with temporal derivatives) … but inferences via F-test
![Page 40: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/40.jpg)
Overview
1. Block/epoch vs. event-related fMRI
2. (Dis)Advantages of efMRI
3. GLM: Convolution
4. BOLD impulse response
5. Temporal Basis Functions
6. Timing Issues
7. Design Optimisation – “Efficiency”
![Page 41: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/41.jpg)
Design efficiency
• HRF can be viewed as a filter (Josephs & Henson, 1999)
• We want to maximise the signal passed by this filter
• Dominant frequency of canonical HRF is ~0.04 Hz
➡ The most efficient design is a sinusoidal modulation of neural activity with period ~24s(e.g., boxcar with 12s on/ 12s off)
![Page 42: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/42.jpg)
⊗ =
× =
A very “efficient” design!
Stimulus (“Neural”) HRF Predicted Data
Sinusoidal modulation, f = 1/33
![Page 43: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/43.jpg)
=
=
Blocked-epoch (with small SOA) quite “efficient”
⊗
×
Blocked, epoch = 20 sec
Stimulus (“Neural”) HRF Predicted Data
![Page 44: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/44.jpg)
× =
⊗
“Effective HRF” (after highpass filtering)(Josephs & Henson, 1999)
Very ineffective: Don’t have long (>60s) blocks!
=
Blocked (80s), SOAmin=4s, highpass filter = 1/120s
Stimulus (“Neural”) HRF Predicted Data
![Page 45: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/45.jpg)
⊗ =
× =
Randomised design spreads power over frequencies
Stimulus (“Neural”) HRF Predicted Data
Randomised, SOAmin=4s, highpass filter = 1/120s
![Page 46: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/46.jpg)
Design efficiency
• T-statistic for a given contrast: T = cTb / var(cTb)
• For maximum T, we want maximum precision and hence minimum standard error of contrast estimates (var(cTb))
• Var(cTb) = sqrt(σ2cT(XTX)-1c) (i.i.d)
• If we assume that noise variance (σ2) is unaffected by changes in X, then our precision for given parameters is proportional to the design efficiency: e(c,X) = { cT (XTX)-1 c }-1
➡ We can influence e (a priori) by the spacing and sequencing of epochs/events in our design matrix
➡ e is specific for a given contrast!
![Page 47: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/47.jpg)
Blocked designs most efficient! (with small SOAmin)
Design efficiency: Trial spacing
• Design parametrised by:
- SOAmin Minimum SOA
- p(t) Probability of event at each SOAmin
• Deterministic p(t)=1 iff t=nSOAmin
• Stationary stochastic p(t)=constant
• Dynamic stochasticp(t) varies (e.g., blocked)
![Page 48: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/48.jpg)
0
22.5
45.0
67.5
90.0
Block Dyn stoch Randomised
Design efficiency: Trial spacing
3 sessions with 128 scansFaces, scrambled facesSOA always 2.97 sCycle length 24 s
e
• However, block designs are often not advisable due to interpretative difficulties (see before)
• Event trains may then be constructed by modulating the event probabilities in a dynamic stochastic fashion
• This can result in intermediate levels of efficiency
![Page 49: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/49.jpg)
Differential Effect (A-B)
Common Effect (A+B)
Design efficiency: Trial sequencing
• Design parametrised by: SOAmin Minimum SOA
pi(h) Probability of event-type i given history h of last m events
• With n event-types pi(h) is a n x n Transition Matrix
• Example: Randomised AB
A B A 0.5 0.5
B 0.5 0.5
=> ABBBABAABABAAA...
![Page 50: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/50.jpg)
Alternating (A-B)
Permuted (A-B)
• Example: Permuted AB
A B AA 0 1 AB 0.5 0.5 BA 0.5 0.5BB 1 0
=> ABBAABABABBA...
Design efficiency: Trial sequencing
• Example: Alternating AB
A B A 0 1 B 1 0
=> ABABABABABAB...
![Page 51: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/51.jpg)
Null Events (A+B)
Null Events (A-B)
Design efficiency: Trial sequencing
• Example: Null events
A B A 0.33 0.33B 0.33 0.33
=> AB-BAA--B---ABB...
• Efficient for differential and main effects at short SOA
• Equivalent to stochastic SOA (Null Event like third unmodelled event-type)
![Page 52: ruff spm efMRI fil 2011 - fil.ion.ucl.ac.uk · SPM{F} Timing issues: Slice Timing “Slice-timing Problem”: ‣ Slices acquired at different times, yet model is the same for all](https://reader034.vdocuments.site/reader034/viewer/2022050717/5e14f5b99527280a1b5715b9/html5/thumbnails/52.jpg)
Design efficiency: Conclusions
‣ Optimal design for one contrast may not be optimal for another
‣ Blocked designs generally most efficient (with short SOAs, given optimal block length is not exceeded)
‣ However, psychological efficiency often dictates intermixed designs, and often also sets limits on SOAs
‣ With randomised designs, optimal SOA for differential effect (A-B) is minimal SOA (>2 seconds, and assuming no saturation), whereas optimal SOA for main effect (A+B) is 16-20s
‣ Inclusion of null events improves efficiency for main effect at short SOAs (at cost of efficiency for differential effects)
‣ If order constrained, intermediate SOAs (5-20s) can be optimal
‣ If SOA constrained, pseudorandomised designs can be optimal (but may introduce context-sensitivity)