中央大學認知神經科學研究所 2015 認知神經科學巡迴工作 坊 FMRI 實驗現場 2 3 Chouinard P A Large ME Chang E C amp Goodale M A 2009 NeuroImage ID: 1048181
Download Presentation The PPT/PDF document "FMRI 的實驗設計 講師：張智宏 ..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
1. FMRI的實驗設計講師：張智宏 副教授中央大學認知神經科學研究所2015 認知神經科學巡迴工作坊
2. FMRI實驗現場2
3. 3Chouinard, P. A., Large, M.-E., Chang, E. C., & Goodale, M. A. (2009). NeuroImage
4. 可能想使用腦造影工具的理由我想看高僧入定時的腦部活動和一般人有什麼不一樣？我看看進行X@#$心理歷程時，有哪些腦區在活動。4
5. 使用腦造影技術前，提醒自己：「如果腦造影結果是答案，那麼研究的問題是什麼？」Stephen Kosslyn (1999)「貴重儀器無助於設計不良的實驗。」Louis Sokoloff對實驗結果有合乎理論脈絡的預期，能回達理論上有趣、尚未解決的問題；避免進行釣魚式的腦造影實驗。能掌握實驗設計以及資料分析概念的研究者，比較有可能生產出有意義的實驗。5
6. 實驗基本要素獨變項(Independent variable, IV)Aspects of the experimental design that are intentionally manipulated and that are hypothesized to cause changes in DVConditions or levelsAt least two conditions/levels for an IV依變項(Dependent variable, DV)Quantities that are measured to evaluate the effect of IVRT, accuracy, trajectory, … etc.ERP, fMRI, MEG6
7. FMRI實驗的基本目的7
8. FMRI 實驗術語ConditionsTrialsEvents8
9. 設計的概念與方法層面Conceptual designHow to design proper tasks to measure the mental process of interest?Methodological designHow to construct task paradigms to optimize the efficiency and power to measure the effects of interest, given multiple constraints in FMRI environment?9
10. 設計之概念層面Categorical designs Subtraction Pure insertion, evoked / differential responses ConjunctionTesting multiple hypotheses Parametric designs LinearAdaptation, cognitive dimensionsNonlinearPolynomial expansions, neurometric functions 10
11. Categorical DesignAim: Neuronal structures underlying a single process P? Procedure Contrast[Task with P] – [control task without P ] = P The critical assumption of „pure insertion“ 11
12. Example: Cognitive subtraction [Task with P] – [task without P ] = P 12
13. Subtraction Logic: Brain Imaging ExampleHypothesis : Some areas of the brain are specialized for perceiving objectsSimplest design: Compare pictures of objects vs. a control stimulus that is not an objectminus= object perceptionseeingpictureslikeseeingpictureslikeMalach et al., 1995, PNAS13
14. Objects > TexturesMalach et al., 1995, PNASLateralOccipitalComplex(LOC)14
15. FMRI Subtraction-=15
16. Parametric DesignEmploys continuous variation in a stimulus/task parameterworking memory load, stimulus contrastInference:Modulation of activity reflects sensitivity to the modulated parameterCan demonstrate more than “where is the activation”: instead, how does this region compute variable XMay make control task unnecessary16
17. Parametric DesignPossible tests for such parametric relationLinear Nonlinear: Quadratic/cubic/etc. (polynomial expansion) Model-based (e.g. predictions from learning models) 17IV LevelBOLD Change
18. Model-based FMRI18
19. Boynton et al. (1996)19
20. Methodological DesignsBlocked designsEvent-related designs20
21. Detection vs. EstimationDetection: determination of whether activity of a given voxel (or region) changes in response to the experimental manipulation“which voxel?”Definitions modified from: Huettel, Song & McCarthy, 2004, Functional Magnetic Resonance Imaging% Signal Change0Time (sec)048121Estimation: measurement of the time course within an active voxel in response to the experimental manipulation“How does signal change in a voxel?”21
22. Design TypesBlockDesignSlow ERDesignRapidCounterbalancedER DesignRapidJittered ERDesignMixedDesign= null trial (nothing happens)= trial of one type (e.g., face image)= trial of another type (e.g., place image)22
23. Block Designs23B1B2Alternating DesignInterleaving null-task blocks
24. Block DesignsEarly Assumption: Because the hemodynamic response delays and blurs the response to activation, the temporal resolution of fMRI is limited.= trial of one type (e.g., face image)= trial of another type (e.g., place image)BlockDesignPositive BOLD responseInitialDipOvershootPost-stimulusUndershoot0123BOLD Response(% signal change)TimeStimulus24
25. First fMRI Results with a Block Design25Kwong et al. (1992) PNAS
26. Advantages and DisadvantagesHigh detection powerTrade-off of block lengthLong blockLarger differences between conditions Short blockAvoid confounding with low frequency scanner drift or subject state (like being bored)26
27. 27
28. Blocked Design使用建議Length of a blockBlock length at hemodynamic response duration (10~15 s)Equivalent for conditions or combination of conditions to be comparedA - B T(A) = T(B)A + B – C T(A) + T(B) = T(C)Evoking the same mental process throughout a block28
29. Event-related Designs29Slow ERDesignRapidCounterbalancedER DesignRapidJittered ERDesign
30. Slow Event-Related DesignsSlow ERDesign30
31. Periodic (Slow) ER DesignFixed and long ISIUsually > 15sEach event evokes a complete hemodynamic response, and corresponding BOLD are selectively averaged.Inefficient 31
32. First fMRI Results with an Event-Related Design32Blamire et al. (1992) PNAS
33. Effects of ISI on ER-FMRI Activation33Bandettinni & Cox (2000)
34. Optimal Constant ISIBrief (< 2 sec) stimuli:optimal trial spacing = 12 secFor longer stimuli:optimal trial spacing = 8 + 2*stimulus duration secEffective loss in power of event related design:= -35%i.e., for 6 minutes of block design, run ~9 min ER designSource: Bandettini et al., 200034
35. 實驗設計效能Efficiency of DesignRelative measure of desirability of an estimator or experiment designProportional to power: higher efficient design more likely detects activationsInvolves comparisons of potentially infinite possibilities/procedures35
36. “Do You Wanna Go Faster?”Yes, but we have to test assumptions regarding linearity of BOLD signal firstRapidJittered ERDesignMixedDesignRapidCounterbalancedER Design36
37. BOLD response的線性程度Source: Dale & Buckner, 1997Linearity:“Do things add up?”red = 2 - 1green = 3 - 2Sync each trial response to start of trialNot quite linear but good enough!37
38. Linearity is okay for events every ~4+ s38
39. 快速隨機變動事件相關設計Rapid Jittered ER DesignRapidJittered ERDesign39= trial of one type (e.g., face image)= trial of another type (e.g., place image)
40. BOLD Overlap With Regular Trial SpacingNeuronal activity from TWO event types with constant ITIPartial tetanus BOLD activity from two event typesSlide from Matt Brown40
41. 隨機變動下的BOLD OverlapNeuronal activity from closely-spaced, jittered eventsBOLD activity from closely-spaced, jittered eventsSlide from Matt Brown41
42. General Linear Model42The modelNormal equationAssuming this is invertible
43. Hypothesis Testing43df: Coefficient of Efficiency
44. Why Is Jittered ISI More Efficient?44Eff(β1) = 1/.5632 = 1.76Eff(β2) = 1/.5703 = 1.75Eff(β1) = 1/.3606 = 2.77Eff(β2) = 1/.4640 = 2.16
45. 為什麼要變動 ISI？Without jittering predictors from different trial types are strongly anticorrelatedAs we know, the GLM doesn’t do so well when predictors are correlated (or anticorrelated)增加BOLD signal中可被實驗解釋的變異程度When pink is on, yellow is off pink and yellow are anticorrelatedIncludes cases when both pink and yellow are off less anticorrelation45
46. Algorithms for Picking Efficient DesignsOptseq2http://surfer.nmr.mgh.harvard.edu/optseq/46
47. Algorithms for Picking Efficient DesignsGenetic Algorithmshttp://wagerlab.colorado.edu/tools47
48. FMRI實驗設計實用建議作業要引發與研究問題相關之心智歷程最大化從每位受試者收集的資料量在受試者能忍受的前提下儘可能多收受試者在計畫經費允許的前提下48
49. FMRI實驗設計實用建議選擇可以讓心智歷程產生最大改變的作業情況與時序安排儘量減少事件之間BOLD的相關性Jitter or slow ER建立作業行為表現與BOLD訊號間的相關性49
50. 比較好的使用腦造影工具的理由實例高僧因為長期練習入定，其注意力控制能力可能優於常人。預期其注意力網路活化程度，比一般人更高、區域更集中、區域間連結更緊密。某種心理歷程可能之神經網路為A+B+C…，預期可利用作業TA, TB, TC, TD之間的對比，分別辨識出A, B, and C.50
51. 參考資料來源：線上Duke BIAChttp://www.biac.duke.edu/education/courses/fall08/fmri/Dr. Jody Cuhlam’s fMRI for newbieshttp://culhamlab.ssc.uwo.ca/fmri4newbies/Tutorials.htmlU of Michigan fMRI training coursehttp://sitemaker.umich.edu/fmri.training.course/2012_lecture_notes51
52. 教科書推薦52Poldrack et al. (2012)Huettel et al. (2014)Linquist & Wager (2015)
53. 53Questions?
54. But…54
55. Other DifferencesIs subtraction logic valid here?What else could differ between objects and textures?Objects > Texturesobject shapesirregular shapesfamiliaritynamabilitySource: Dr. Jody Culham’s fMRI for newbies55visual features (e.g., brightness, contrast, etc.)actabilityattention-grabbing
56. Other SubtractionsLateral Occipital ComplexVisual Cortex (V1)Malach et al., 1995, PNAS>>>Grill-Spector et al., 1998, NeuronKourtzi & Kanwisher, 2000, J NeurosciSource: Dr. Jody Culham’s fMRI for newbies56
57. Linearity is okay for events every ~4+ s57