1. MRN fMRI Course Lecture 3.4 (1h): Group ICA Vince D. Calhoun, Ph.D. Chief Technology Officer & Director, Image Analysis & MR Research The Mind Research Network Associate Professor, Electrical and Computer Engineering, Neurosciences, and Computer Science The University of New Mexico

2. 2010 MRN fMRI Course 2

3. 3 Using ICA to analyze fMRI data of multiple subjects raises some questions:Using ICA to analyze fMRI data of multiple subjects raises some questions: How are components to be combined across subjects?How are components to be combined across subjects? How should the final results be thresholded and/or presented?How should the final results be thresholded and/or presented?

4. 2010 MRN fMRI Course 4 Group ICA Sub 1 Sub N ICA ICA ?

5. 2010 MRN fMRI Course 5 Group ICA Approaches

6. 2010 MRN fMRI Course 6 Approach 1 Separate ICA analysis for each subject [V. D. Calhoun, T. Adali, V. McGinty, J. J. Pekar, T. Watson, and G. D. Pearlson, "FMRI Activation In A Visual- Perception Task: Network Of Areas Detected Using The General Linear Model And Independent Components Analysis," NeuroImage, vol. 14, pp. 1080-1088, 2001.]Separate ICA analysis for each subject [V. D. Calhoun, T. Adali, V. McGinty, J. J. Pekar, T. Watson, and G. D. Pearlson, "FMRI Activation In A Visual- Perception Task: Network Of Areas Detected Using The General Linear Model And Independent Components Analysis," NeuroImage, vol. 14, pp. 1080-1088, 2001.] Must select which components to compare between the individualsMust select which components to compare between the individuals Sub 1 Sub N ICA ICA ?

7. 2010 MRN fMRI Course 7 Example … 15 “events” Press buttons (1-4) to indicate choice 1234 Time (seconds) 015.4 31.547.0300

8. 2010 MRN fMRI Course 8 SPM Results N=10 P<0.05 corrected SPM revealed a large network of areas including: frontal eye fieldsfrontal eye fields supplementary motor areassupplementary motor areas primary visualprimary visual visual associationvisual association basal gangliabasal ganglia thalamic, and anthalamic, and an (unexpectedly) large cerebellar activation(unexpectedly) large cerebellar activation bilateral inferior parietal regions were deactivated (not shown)bilateral inferior parietal regions were deactivated (not shown)

9. 2010 MRN fMRI Course 9 ICA Results N=10Z>3.1 ICA revealed a large network of similar areas including: frontal eye fields (blue)frontal eye fields (blue) supplementary motor areas (green w/ outline)supplementary motor areas (green w/ outline) primary visual (red)primary visual (red) visual association (red)visual association (red) thalamic (red)thalamic (red) basal ganglia (green w/ outline)basal ganglia (green w/ outline) a large cerebellar activation (red)a large cerebellar activation (red) bilateral inferior parietal deactivations (not shown)bilateral inferior parietal deactivations (not shown) ICA also revealed areas not identified by SPM including: primary motor (green)primary motor (green) frontal regions anterior to the frontal eye fields (blue)frontal regions anterior to the frontal eye fields (blue) superior parietal regions (blue)superior parietal regions (blue)

10. 2010 MRN fMRI Course 10 ICA: Single Subject The ICA maps from one subject for the visual and basal ganglia components are depicted along with their time courses (basal ganglia in green and visual in pink) Note that the visual time course precedes the motor time course

11. 2010 MRN fMRI Course 11 Event-Averaged Time Courses Time courses from selected voxels in the raw data (a) and time courses produced by the ICA method (b).Time courses from selected voxels in the raw data (a) and time courses produced by the ICA method (b). In all cases the time courses are event-averaged (according to when the figure was presented) within each participant and then averaged across all ten participants.In all cases the time courses are event-averaged (according to when the figure was presented) within each participant and then averaged across all ten participants. Voxels from the raw data were selected by choosing a local maximum in the activation map and averaging the two surrounding voxels in each direction.Voxels from the raw data were selected by choosing a local maximum in the activation map and averaging the two surrounding voxels in each direction. Dashed lines indicate the standard error of the mean.Dashed lines indicate the standard error of the mean.

12. 2010 MRN fMRI Course 12 Approach 2 Group ICA (stacking images)Group ICA (stacking images) [V. D. Calhoun, T. Adali, G. D. Pearlson, and J. J. Pekar, "A Method for Making Group Inferences From Functional MRI Data Using Independent Component Analysis," Hum. Brain Map., vol. 14, pp. 140-151, 2001.][V. D. Calhoun, T. Adali, G. D. Pearlson, and J. J. Pekar, "A Method for Making Group Inferences From Functional MRI Data Using Independent Component Analysis," Hum. Brain Map., vol. 14, pp. 140-151, 2001.] [V. J. Schmithorst and S. K. Holland, "Comparison of Three Methods for Generating Group Statistical Inferences From Independent Component Analysis of Functional Magnetic Resonance Imaging Data," J. Magn Reson. Imaging, vol. 19, pp. 365-368, 2004.][V. J. Schmithorst and S. K. Holland, "Comparison of Three Methods for Generating Group Statistical Inferences From Independent Component Analysis of Functional Magnetic Resonance Imaging Data," J. Magn Reson. Imaging, vol. 19, pp. 365-368, 2004.] Components and time courses can be directly comparedComponents and time courses can be directly compared Sub 1 Sub N ICA Sub 1 Sub N

13. 2010 MRN fMRI Course 13 Group ICA X Subject 1 Subject N Data A S_aggICA A1A1A1A1 ANANANAN Subject i Back-reconstruction AiAiAiAi SiSiSiSi

14. 2010 MRN fMRI Course 14 Simulation Nine simulated source maps and time courses were generated, followed by an ICA estimation. The red lines indicate the t <4.5 boundaries

15. 2010 MRN fMRI Course 15 Are the data separable? (Simulation) A natural concern is whether the back-reconstructed maps from individual subjects will be influenced by the other subjects in the group analysisA natural concern is whether the back-reconstructed maps from individual subjects will be influenced by the other subjects in the group analysis This simulation was performed in which one of the nine “subjects” had a structured, source #2 map (whereas all of the nine “subjects” had a similar, source #1 map).This simulation was performed in which one of the nine “subjects” had a structured, source #2 map (whereas all of the nine “subjects” had a similar, source #1 map). As one can see, in this example, the back-reconstructed ICA maps are very close to the individual maps and there appears to be little to no influence between subjectsAs one can see, in this example, the back-reconstructed ICA maps are very close to the individual maps and there appears to be little to no influence between subjects

16. 2010 MRN fMRI Course 16 The Stationarity Assumption The ICA estimation requires the data to be stationary across subjectsThe ICA estimation requires the data to be stationary across subjects Some signals in the data (e.g. physiologic noise) will most likely *not* be stationarySome signals in the data (e.g. physiologic noise) will most likely *not* be stationary However it is reasonable to assume the signal of interest (fMRI activation) will be stationaryHowever it is reasonable to assume the signal of interest (fMRI activation) will be stationary A simulation was performed to examine how non-stationary sources would affect the resultsA simulation was performed to examine how non-stationary sources would affect the results One stationary signal (fMRI activation) and one non-stationary signal were simulated for a five- subject analysisOne stationary signal (fMRI activation) and one non-stationary signal were simulated for a five- subject analysis The ICA results reveal that the fMRI activation is preservedThe ICA results reveal that the fMRI activation is preserved ICA results Stationary source S common to all five “subjects” source #2 source #1 Sources S1-S5 differing across the five “subjects” S1S2 S3S4S5 S

17. 2010 MRN fMRI Course Evaluation of Group ICA Methods 17 E. Erhardt, S. Rachakonda, E. Bedrick, T. Adali, and V. D. Calhoun, "Comparison of multi-subject ICA methods for analysis of fMRI data," in Proc. HBM, Barcelona, Spain, 2010.

18. 2010 MRN fMRI Course Comparison of multi-subject ICA methods for analysis of fMRI data 18 E. Erhardt, S. Rachakonda, E. Bedrick, T. Adali, and V. D. Calhoun, "Comparison of multi-subject ICA methods for analysis of fMRI data," in Proc. HBM, Barcelona, Spain, 2010.

19. 2010 MRN fMRI Course Comparison of multi-subject ICA methods for analysis of fMRI data 19 E. Erhardt, S. Rachakonda, E. Bedrick, T. Adali, and V. D. Calhoun, "Comparison of multi-subject ICA methods for analysis of fMRI data," in Proc. HBM, Barcelona, Spain, 2010. GICA3 STR DIFF

20. 2010 MRN fMRI Course Default Mode Group Maps 20 E. Erhardt, S. Rachakonda, E. Bedrick, T. Adali, and V. D. Calhoun, "Comparison of multi-subject ICA methods for analysis of fMRI data," in Proc. HBM, Barcelona, Spain, 2010. GICA3STR

21. 2010 MRN fMRI Course 21 Methods Scan Parameters 9 slice Single-shot EPI FOV = 24cm, 64x64 TR=1s, TE=40ms Thickness = 5/.5 mm 360 volumes acquired Preprocessing Timing correction Motion correction Normalization Smoothing ICA An ICA estimation was performed on each of the nine subjects Data were first reduced from 360 to 25 using PCA, the data were concatenated and reduced a second time from 225 to 20 using PCA An ICA estimation was performed after which single subject maps and time courses were calculated Group averaged maps were thresholded at t<4.5, colorized, and overlaid onto an EPI scan for visualization 090180270360 Left Right t (secs)  

22. 2010 MRN fMRI Course 22 Are the data separable? (fMRI experiment) The same slice from nine subjects when the right (red) and left (blue) visual fields were stimulated, (a) analyzed via linear modeling (LM), (b) back-reconstructed from a group ICA analysis, or (c) calculated from an ICA analysis performed on each subject separately. A transiently task-related component is depicted in green.The same slice from nine subjects when the right (red) and left (blue) visual fields were stimulated, (a) analyzed via linear modeling (LM), (b) back-reconstructed from a group ICA analysis, or (c) calculated from an ICA analysis performed on each subject separately. A transiently task-related component is depicted in green. The results between the two ICA methods appear quite similar and match well with the LM results as well (note that there may be small differences due to different initial conditions for the ICA estimation)The results between the two ICA methods appear quite similar and match well with the LM results as well (note that there may be small differences due to different initial conditions for the ICA estimation)

23. 2010 MRN fMRI Course 23 Comparison with GLM Approach R L V.D. Calhoun, T. Adali, G.D. Pearlson, and J.J. Pekar, "A Method for Making Group Inferences From Functional MRI Data Using Independent Component Analysis," Hum. Brain Map., vol. 14, pp. 140-151, 2001.

24. 2010 MRN fMRI Course 24 Sorting/Calibrating A ‘second-level’ or group analysis involves taking certain parameters (estimated by ICA) such as the amplitude fit for fMRI regression models, or voxel weights, and testing these within a standard GLM hypothesis-testing frameworkA ‘second-level’ or group analysis involves taking certain parameters (estimated by ICA) such as the amplitude fit for fMRI regression models, or voxel weights, and testing these within a standard GLM hypothesis-testing frameworkComp# R2R2R2R2SubjectReg1Reg210.8111.890.02 22.280.66 100.8110.282.19 20.652.03 40.0171-0.19-0.40 2-0.100.08

25. 2010 MRN fMRI Course Prenormalization 25 E. Allen, E. Erhardt, T. Eichele, A. R. Mayer, and V. D. Calhoun, "Comparison of pre-normalization methods on the accuracy of group ICA results," in Proc. HBM, Barcelona, Spain, 2010. 1) No Normalization (NN), where data is left in its raw intensity units (Calhoun, 2001) 2) Intensity Normalization (IN), which involves voxel- wise division of the time series mean 3) Variance Normalization (VN), voxel-wise z-scoring of the time series (Beckmann, 2004).

26. 2010 MRN fMRI Course 26 Comp# DescriptionCorr OddballRest 1619 A: Default mode 0.9577 119 B: Motor 0.9156 1312 C: Sup parietal 0.9142 106 D: Medial visual 0.8628 127 E: Left lateral frontoparietal 0.8557 142 F: Lateral Visual 0.8170 1713 G: Temporal2 0.8135 811 H: Cerebellum 0.8059 115 I: Temporal1 0.8048 416 J: Frontal 0.7838 24 K: Right lateral frontoparietal 0.8170 5 L: Anterior cingulate 0.035 Result 1: AOD and rest data produced highly similar networks V. D. Calhoun, K. A. Kiehl, and G. D. Pearlson, "Modulation of Temporally Coherent Brain Networks Estimated using ICA at Rest and During Cognitive Tasks," Hum Brain Mapp, vol. 29, pp. 828-838, 2008.

27. 2010 MRN fMRI Course 27 DescriptionTarNov A: Default mode-8.44 (1.4e-9)-5.79 (5.6e-6) B: Motor4.62 (2.3e-4)1.11 (1.0) C: Sup parietal2.51 (8.9e-2)-3.50 (6.5e-3) D: Medial visual1.09 (1.0)0.12 (1.0) E: Left lateral frontoparietal2.41 (1.1e-1)1.21 (1.0) F: Lateral Visual-4.34 (5.4e-4)-3.92 (1.9e-3) G: Temporal210.29 (6.2e-12)7.76 (1.1e-8) H: Cerebellum4.09 (1.1e-3)-2.59 (7.4e-2) I: Temporal113.67 (1.2e-15)9.30 (1.1e-10) J: Frontal-2.55 (8.1e-2)-3.28 (1.2e-2) K: Right lateral frontoparietal-12.00 (6.3e-15)-3.89 (2.1e-3) Result 2: Though similar TCNs were identified for AOD and rest, spatial and temporal task modulation was induced V. D. Calhoun, K. A. Kiehl, and G. D. Pearlson, "Modulation of Temporally Coherent Brain Networks Estimated using ICA at Rest and During Cognitive Tasks," Hum Brain Mapp, vol. 29, pp. 828-838, 2008.

28. 2010 MRN fMRI Course 28 Example of spatial sorting

29. 2010 MRN fMRI Course 29 Example 1: ‘Default Mode’ Mask Using wfu pickatlas to define mask using regions reported in Rachle 2001 paperUsing wfu pickatlas to define mask using regions reported in Rachle 2001 paper Posterior parietal cortex BA7Posterior parietal cortex BA7 Occipitoparietal junction BA 39Occipitoparietal junction BA 39 PrecuneusPrecuneus Posterior cingulatePosterior cingulate Frontal Pole BA 10Frontal Pole BA 10 Smooth in SPM with same kernel used on fMRI dataSmooth in SPM with same kernel used on fMRI data Sort in GIFT using spatial sortingSort in GIFT using spatial sorting A.Garrity, G.D.Pearlson, K.McKiernan, D.Lloyd, K.A.Kiehl, and V.D.Calhoun, "Aberrant 'Default Mode' Functional Connectivity in Schizophrenia," to appear Am. J. Psychiatry, 2006.

30. 2010 MRN fMRI Course 30 ICA to identify ‘Default Mode’ Network HealthySchizo Healthy vs Schizo (N=26/26) +Symptoms A.Garrity, G.D.Pearlson, K.McKiernan, D.Lloyd, K.A.Kiehl, and V.D.Calhoun, "Aberrant 'Default Mode' Functional Connectivity in Schizophrenia," to appear Am. J. Psychiatry, 2006.

31. 2010 MRN fMRI Course 31 Spatial Sorting: Example 2 Classification of SchizophreniaClassification of Schizophrenia Mapping the brain via intrinsic connectivityMapping the brain via intrinsic connectivity Patients Controls

32. 2010 MRN fMRI Course 32 Robustness of ‘modes’

33. 2010 MRN fMRI Course 33 The Challenge Accurate classification requires single-subject accuracy -> very stringent requirement!Accurate classification requires single-subject accuracy -> very stringent requirement! We cannot use knowledge of the diagnosis in the development of the classification algorithmWe cannot use knowledge of the diagnosis in the development of the classification algorithm

34. 2010 MRN fMRI Course 34 Temporal Lobe Synchrony Supervised ClassificationSupervised Classification Step 1: Select Training GroupStep 1: Select Training Group Step 2: Use ICA to extract temporal lobe mapsStep 2: Use ICA to extract temporal lobe maps Step 3: Compute within-group mean imagesStep 3: Compute within-group mean images Step 4: Subtract the mean imagesStep 4: Subtract the mean images Step 5: Set a positive and negative thresholdStep 5: Set a positive and negative threshold HC 1 HC N ICA Sz 1 Sz N … … Calhoun VD, Kiehl KA, Liddle PF, Pearlson GD: “Aberrant Localization of Synchronous Hemodynamic Activity in Auditory Cortex Reliably Characterizes Schizophrenia”. Biol Psychiatry 2004; 55842-849

35. 2010 MRN fMRI Course 35 Temporal Lobe Synchrony in Schizophrenia

36. 2010 MRN fMRI Course 36 Temporal Lobe Synchrony in Schizophrenia Step 6: Form classification measure (average the values within each boundary and subtract)Step 6: Form classification measure (average the values within each boundary and subtract) Step 7: Optimize group discrimination (using a sensible error metric)Step 7: Optimize group discrimination (using a sensible error metric) Step 8: Apply classification to new dataStep 8: Apply classification to new data Calhoun VD, Kiehl KA, Liddle PF, Pearlson GD: “Aberrant Localization of Synchronous Hemodynamic Activity in Auditory Cortex Reliably Characterizes Schizophrenia”. Biol Psychiatry 2004; 55842-849

37. 2010 MRN fMRI Course 37 Temporal Sorting: fBIRN SIRP Task MethodsMethods Subjects & TaskSubjects & Task 28 subjects (14 HC/14 SZ) across two sites28 subjects (14 HC/14 SZ) across two sites Three runs of SIRP task preprocessed with SPM2Three runs of SIRP task preprocessed with SPM2 ICA AnalysisICA Analysis All data entered into group ICA analysis in GIFTAll data entered into group ICA analysis in GIFT ICA time course and image reconstructed for each subject, session, and componentICA time course and image reconstructed for each subject, session, and component Images: sessions averaged together creating single image for each subject and componentImages: sessions averaged together creating single image for each subject and component Time courses: SPM SIRP model regressed against ICA time courseTime courses: SPM SIRP model regressed against ICA time course Statistical Analysis:Statistical Analysis: Images: all subjects entered into voxelwise 1-sample t-test in SPM2 and thresholded at t=4.5Images: all subjects entered into voxelwise 1-sample t-test in SPM2 and thresholded at t=4.5 Time courses: Goodness of fit to SPM SIRP model computed, beta weights for load 1, 3, 5 entered into Group x Load ANOVATime courses: Goodness of fit to SPM SIRP model computed, beta weights for load 1, 3, 5 entered into Group x Load ANOVA fBIRN Phase II Data: www.nbirn.net; NCRR (NIH), 5 MOI RR 000827 (2002-2006) and 1 U24 RR0219921 (2006 onwards) www.nbirn.net

38. 2010 MRN fMRI Course 38 Component 1: Bilateral Frontal/Parietal fBIRN Phase II Data: www.nbirn.net; NCRR (NIH), 5 MOI RR 000827 (2002-2006) and 1 U24 RR0219921 (2006 onwards) www.nbirn.net

39. 2010 MRN fMRI Course 39 Component 2: Right Frontal, Left Parietal, Post. Cing. fBIRN Phase II Data: www.nbirn.net; NCRR (NIH), 5 MOI RR 000827 (2002-2006) and 1 U24 RR0219921 (2006 onwards) www.nbirn.net

40. 2010 MRN fMRI Course 40 Component 3: Temporal Lobe fBIRN Phase II Data: www.nbirn.net; NCRR (NIH), 5 MOI RR 000827 (2002-2006) and 1 U24 RR0219921 (2006 onwards) www.nbirn.net

41. 2010 MRN fMRI Course 41 Example 2: Simulated Driving Paradigm

42. 2010 MRN fMRI Course 42 Previous Work Walter, 2001.Walter, 2001. Driving Watching “our study suggests that the main ideas of cognitive psychology used in the design of cars, in the planning of respective behavioral experiments on driving, as well as in traffic related political decision making (i.e. laws on what drivers are supposed to do and not to do during driving) may be inadequate, as it suggests a general limited capacity model of the psyche of the driver which is not supported by our results. If driving deactivates rather than activates a number of brain regions the quests for the adequate design of the man-machine interface as well as for what the driver should and should not do during driving is still widely open.” “Our results suggest that simulated driving engages mainly areas concerned with perceptual- motor integration and does not engage areas associated with higher cognitive functions.”

43. 2010 MRN fMRI Course 43 Baseline Simulated Driving Results Higher Order Visual/Motor: Increases during driving; less during watching. Low Order Visual: Increases during driving; less during watching. Motor control: Increases only during driving. Vigilance: Decreases only during driving; amount proportional to speed. Error Monitoring & Inhibition: Decreases only during driving; rate proportional to speed. Visual Monitoring: Increases during epoch transitions. N=12 * Drive Watch V. D. Calhoun, J. J. Pekar, V. B. McGinty, T. Adali, T. D. Watson, and G. D. Pearlson, "Different Activation Dynamics in Multiple Neural Systems During Simulated Driving," Hum. Brain Map., vol. 16, pp. 158-167, 2002.

44. 2010 MRN fMRI Course 44 SPM Results Calhoun, V. D., Pekar, J. J., and Pearlson, G. D. “Alcohol Intoxication Effects on Simulated Driving: Exploring Alcohol- Dose Effects on Brain Activation Using Functional MRI”. Neuropsychopharmacology 2004.

45. 2010 MRN fMRI Course 45 Interpretation of Results

46. 2010 MRN fMRI Course 46 Functional Network Connectivity (between groups) B: Parietal C: L. & M. Visual Cortical Areas D: Frontal Temporal Parietal E: Frontal Parietal Subcortical F: Frontal G: Temporal A: Default Key: : ρ patient > ρ control : ρ control > ρ patient

47. 2010 MRN fMRI Course 47 FNC Software

48. 2010 MRN fMRI Course 48 Fusion ICA Toolbox (FIT) http://icatb.sourceforge.net Funded by NIH 1 R01 EB 005846 500+ unique downloads

49. 2010 MRN fMRI Course 49 FMRI Snapshots (movie) Calhoun, V.D., Pearlson, G.D., and Kiehl, K.A. (2006). Neuronal Chronometry of Target Detection: Fusion of Hemodynamic and Event-related Potential Data. NeuroImage 30, 544-553.

50. 2010 MRN fMRI Course 50 SNPsGenes rs1800545rs7412rs1128503rs6578993rs1045642rs2278718rs4784642rs521674ADRA2AAPOEABCB1THABCB1MDH1GNAO1ADRA2A 0.55 Target Stimuli Novel StimuliSNPsGenesrs1800545rs7412rs6578993rs2278718rs1128503rs429358rs3813065rs4121817rs521674ADRA2AAPOETHMDH1ABCB1APOEPIK3C3PIK3C3ADRA2A 0.47

51. 2010 MRN fMRI Course 51 Demo 3 subject ICA3 subject ICA SortingSorting Component Explorer (split time courses, event-related average)Component Explorer (split time courses, event-related average) Orthogonal ViewerOrthogonal Viewer Composite ViewerComposite Viewer Examine Regression ParametersExamine Regression Parameters Taking Images/Timecourses from GIFT to SPMTaking Images/Timecourses from GIFT to SPM

52. 2010 MRN fMRI Course 52

53. 2010 MRN fMRI Course Comparison of multi-subject ICA methods for analysis of fMRI data 53 E. Erhardt, S. Rachakonda, E. Bedrick, T. Adali, and V. D. Calhoun, "Comparison of multi-subject ICA methods for analysis of fMRI data," in Proc. HBM, Barcelona, Spain, 2010.