General Linear Model. Modeling the Brain? ICA vs PCA

Transcription

1 Group ICA Introduction and Review of Previous Work Vince D. Calhoun, Ph.D. Executive Science Officer & Director, Image Analysis & MR Research The Mind Research Network Professor, Electrical and Computer Engineering (primary), Neurosciences, Psychiatry and Computer Science The University of New Mexico Results Modeling Modeling the Brain? Discussion From Science with a Smile by Subramanian Raman All models are wrong, but some are useful! All models are wrong. G.E. Box (1976) quoted by Marks Nester in, An applied statistician s creed, Applied Statistics, 45(4): , I believe in ignorance-based methods because humans have a lot of ignorance and we should play to our strong suit. Eric Lander, Whitehead Institute, M.I.T. 2 ICA vs PCA { 1 2} = { 1} { 2} ( 1 2) = ( 1) ( 2) E{ h( y1) h( y2) } E{ h( y1) } E{ h( y2) } Uncorrelated: E y y E y E y Independent: p y, y p y p y = 1. Model (1 or more Regressors) 2. Data General Linear Model or xi ( j) y( j) PCA finds directions of maximal variance (using second order statistics) ICA finds directions which maximize independence (using higher order statistics) 3. Fitting the Model to the Data at each voxel ( ) = ˆ β + ˆ β ( ) + ( ) 0 M y j x j e j i= 1 i i Regression Results 3 4

2 Time Voxels Data(X) = G General Linear Model (GLM) Design matrix Activation maps Corresponding to columns of G The GLM is by far the most common approach to analyzing fmri data. ˆβ To use this approach, one needs a model for the fmri time course Time courses ICA Example Independent Component Analysis (ICA) Time Voxels Data(X) = ˆ 1 W Mixing matrix Spatially Independent Components Components (C) Time courses In spatial ICA, there is no model for the fmri time course, this is estimated along with the hemodynamic source locations 5 6 ICA Halloween (Un)Mixer! X = A S background Comparison of Different Algorithms Time = candle 1 candle 2 candle 3 Candle out 7 N. Correa, T. Adali, Y. Li, and V. D. Calhoun, "Comparison of Blind Source Separation Algorithms for FMRI Using a New Matlab Toolbox: GIFT," in Proc. ICASSP, Philadelphia, PA, N. Correa, T. Adali, and V. D. Calhoun "Performance of Blind Source Separation Algorithms for fmri Analysis," Mag.Res.Imag.,

3 : Consistency of Infomax A Few Software Packages The ICA:DTU toolbox ( matlab three different ICA algorithms fmri specific with demo data FMRIB Software Library, which includes the ICA tool MELODIC ( melodic/): C FastICA+ Complete Package AnalyzeFMRI ( software.html) R FastICA FMRLAB ( matlab infomax algorithm fmri specific with additional tools ICALAB matlab multiple ICA algorithms not fmri specific although one fmri example included GIFT ( matlab >9 ICA algorithms (more coming) including infomax and fastica Constrained ICA algorithms Visualization tools and sorting options. Sample data and a step-by-step walk through BrainVoyager( Commercial FastICA Complete Package N. Correa, T. Adali, Y. Li, and V. D. Calhoun, "Comparison of Blind Source Separation Algorithms for FMRI Using a New Matlab Toolbox: GIFT," in Proc. ICASSP, Philadelphia, PA, N. Correa, T. Adali, and V. D. Calhoun "Performance of Blind Source Separation Algorithms for fmri Analysis," Mag.Res.Imag., Group ICA [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 , 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 , 2004.] [E. Erhardt, S. Rachakonda, E. Bedrick, T. Adali, and V. D. Calhoun, "Comparison of multi-subject ICA methods for analysis of fmri data," Hum Brain Mapp, In Press] Components and time courses can be directly compared Sub 1 Sub 1 Sub N ICA Sub N Subject 1 Subject N Group ICA Approaches a Combine Single b Temporal c Spatial d Pre-Averaging 5 e Concatenation 3,7,5 Concatenation 6,5 Common Spatial Subject ICA s 1,4 Common Temporal Unique Spatial Unique Temporal Subject 1 }Correlate/Cluster Subject N Time Common Spatial Unique Temporal Voxels : Back reconstruction Time Unique Spatial Common Temporal Voxels Subject 1 }Single subject maps Single subject components* Subject N Subject (avg) GIFT MELODIC Brain Voyager Time Subs Tensor 2,7 Common Spatial Common Temporal Subject Parameter Voxels Subject 1 Subject 1 1) Calhoun VD, Adali T, McGinty V, Pekar JJ, Watson T, Pearlson GD. (2001): fmri Activation In A Visual-Perception Task: Network Of Areas Detected Using The General Linear Model And Independent Component Analysis. NeuroImage 14(5): ) Beckmann CF, Smith SM. (2005): Tensorial extensions of independent component analysis for multisubject FMRI analysis. NeuroImage 25(1): ) Calhoun VD, Adali T, Pearlson GD, Pekar JJ. (2001): A Method for Making Group Inferences from Functional MRI Data Using Independent Component Analysis. Hum.Brain Map. 14(3): ) Esposito F, Scarabino T, Hyvarinen A, Himberg J, Formisano E, Comani S, Tedeschi G, Goebel R, Seifritz E, Di SF. (2005): Independent component analysis of fmri group studies by self-organizing clustering. Neuroimage. 25(1): ) Schmithorst VJ, Holland SK. (2004): Comparison of three methods for generating group statistical inferences from independent component analysis of functional magnetic resonance imaging data. J.Magn Reson.Imaging 19(3): ) Svensen M, Kruggel F, Benali H. (2002): ICA of fmri Group Study Data. NeuroImage 16: ) Guo Y, Giuseppe P. (In Press): A unified framework for group independent component analysis for multi-subject fmri data. NeuroImage. 11 V. D. Calhoun, J. Liu, and T. Adali, "A Review of Group ICA for fmri Data and ICA for Joint Inference of Imaging, Genetic, and ERP data," NeuroImage, vol. 45, pp , 2009, PMC

4 ICA (Forward Estimation) Data PCA PCA ICA Subject 1 : reduction1 reduction2 A 1 = A Group ICA S_agg Back-reconstruction through inversion 1 A i 1 Subject i = S i Evaluation of Group ICA Methods Subject N A N Back-reconstruction through Spatial-temporal (dual) regression 2,3 1) Regress S_agg onto each timepoint of Subject i to generate A i 2) Regress onto each image of Subject i to generate A i S i Iterate steps 1 & 2 until converged 1. V. D. Calhoun et al Hum.Brain Map., vol. 14, pp , V. D. Calhoun et al Neuropsychopharmacology, vol. 29, pp , N. Filippini et al Proc Natl Acad Sci U S A, vol. 106, pp , Apr E. Erhardt, S. Rachakonda, E. Bedrick, T. Adali, and V. D. Calhoun, "Comparison of multi-subject ICA methods for analysis of fmri data," Human Brain Mapping, vol. 12, pp , Mean Simulation Inter-subject Covariation GICA3 Binary Continuous Binary Continuous Co-variate Co-variate Interaction Mean Co-variate Co-variate Interaction Mean STR/DR Binary Continuous Co-variate Co-variate Interaction Default Mode Group Maps Component Number GICA3 STR E. Erhardt, S. Rachakonda, E. Bedrick, T. Adali, and V. D. Calhoun, "Comparison of multi-subject ICA methods for analysis of fmri data," Human Brain Mapping, vol. 12, pp , E. Erhardt, S. Rachakonda, E. Bedrick, T. Adali, and V. D. Calhoun, "Comparison of multi-subject ICA methods for analysis of fmri data," Human Brain Mapping, vol. 12, pp ,

5 Group ICA w/ back-reconstruction provides single subject results Stationary source S common to all five subjects Single-subject variability Sources S1-S5 differing across the five subjects A 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). 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 subjects ICA results The ICA estimation requires the data to be stationary across subjects Some 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 stationary A 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 fivesubject analysis The ICA results reveal that the fmri activation is preserved 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 , 2001 source #1 source # S S1 S2 S3 S4 S5 Single Subject ICA vs Group ICA A simulation toolbox Sub 1 ICA? ICA Sub N Motivation Simulated data facilitate comparative evaluations Creating reasonable simulations can be challenging given the complexities of fmri data SimTB goals Flexible, easy and fast generation of simulated fmri data under an ICA model Ability to answer basic questions regarding the group ICA model E. Erhardt, E. A. Allen, Y. Wei, T. Eichele, and V. D. Calhoun, "SimTB, a simulation toolbox for fmri data under a model of spatiotemporal separability," NeuroImage, Under Review. & OHBM 2011 Abstract #657. E. A. Allen, E. Erhardt, Y. Wei, T. Eichele, and V. D. Calhoun, "Capturing inter-subject variability with group independent component analysis of fmri data: a simulation study," NeuroImage, Under Review. & OHBM 2011 Abstract #690. C nn r i = icrs ic + i = Ri ( gi ) Si + i c= 1 Y g N diag N

6 Overview of simtb Spatial Maps Roughly 30 SMs can be selected through the graphical user interface by clicking on components, or by component number in the batch script For each subject, each selected component can be included or not, be translated and rotated in space, and the spatial spread of the component can be increased or decreased Each SM is normalized to have a peak-to-peak amplitude range of one E. Erhardt, E. A. Allen, Y. Wei, T. Eichele, and V. D. Calhoun, "SimTB, a simulation toolbox for fmri data under a model of spatiotemporal separability," NeuroImage, Under Review. & OHBM 2011 Abstract #657. Moments compared with real fmri data Two paradigm examples Tissue types affect baseline intensity Possible to match simulation to real fmri data x- and y- translation and rotation Subject 1 (bold) has half the motion of subject

7 Overview Motivation A brief review of group ICA A Testing Framework Application to large study Application to disease groups A Simulation Framework & Toolbox (simtb) Pushing the limits of Group ICA Summary Group ICA: Pushed to the Limits Four Experiments Varied component rotation Varied component position Varied component amplitude Varied the temporal correlation among components E. A. Allen, E. Erhardt, Y. Wei, T. Eichele, and V. D. Calhoun, "Capturing inter-subject variability with group independent component analysis of fmri data: a simulation study," NeuroImage, Under Review. & OHBM 2011 Abstract # Rotation Position 27 28

8 Amplitude Temporal Correlation Among Components (FNC) Methods Scan Parameters 9 slice Single-shot EPI FOV = 24cm, 64x64 TR=1s, TE=40ms Right Thickness = 5/.5 mm 360 volumes acquired Preprocessing Left Timing correction t (secs) 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 ! +! +! 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 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) 31 32

9 Comp# R2 Subject Reg1 Reg Group GLM Comparison with GLM in fmri Data Group ICA 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 framework 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 , Comp# R 2 Subject Reg1 Reg Key: : ρ patient > ρ control : ρ control > ρ patient G: Temporal Functional Network Connectivity A: Default B: Parietal Time Voxels Data Subject 1 X Time Components A 1 ICA A S_agg Time Components A i 1 Time Back-reconstruction Voxels Subject i Compoents Voxels S i Subject N A N F: Frontal E: Frontal Parietal Subcortical D: Frontal Temporal Parietal C: L. & M. Visual Cortical Areas Component Images Voxel-wise stats (e.g. one-sample t-test, two-sample t-test, correlation, etc) } } Component images (one per subject) T-statistic Task-modulation (e.g. Fit timecourses to GLM model then test parameter) } } Component Timecourses } Model timecourses Multiple regression fit to ICA timecourses Beta-weights (second level model) T-value Spectra (e.g. power spectra, fractal parameters, etc) Controls-Patients Frequency (Hz) T-Values } Component timecourses } Power spectra group differences Functional Network Connectivity (e.g. inter-component correlation) M. Jafri, G. D. Pearlson, M. Stevens, and V. D. Calhoun, "A Method for Functional Network Connectivity Among Spatially Independent Resting-State Components in Schizophrenia," NeuroImage, vol. 39, pp ,

10 Result 1: AOD and rest data produced highly similar networks Result 2: Though similar TCNs were identified for AOD and rest, spatial and temporal task modulation was induced Comp# Comp# Description Corr Oddball Rest A: Default mode B: Motor C: Sup parietal D: Medial visual E: Left lateral frontoparietal F: Lateral Visual G: Temporal H: Cerebellum I: Temporal J: Frontal K: Right lateral frontoparietal L: Anterior cingulate Description Tar Nov A: Default mode (1.4e-9) (5.6e-6) B: Motor 4.62 (2.3e-4) 1.11 (1.0) C: Sup parietal 2.51 (8.9e-2) (6.5e-3) D: Medial visual 1.09 (1.0) 0.12 (1.0) E: Left lateral frontoparietal 2.41 (1.1e-1) 1.21 (1.0) F: Lateral Visual (5.4e-4) (1.9e-3) G: Temporal (6.2e-12) 7.76 (1.1e-8) H: Cerebellum 4.09 (1.1e-3) (7.4e-2) I: Temporal (1.2e-15) 9.30 (1.1e-10) J: Frontal (8.1e-2) (1.2e-2) K: Right lateral frontoparietal (6.3e-15) (2.1e-3) 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 , 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 , Collaborative Informatics & Neuroimaging Suite (COINS) Variability/in/a/Healthy/Sample/(N=603)/ h"p://coins.mrn.org/ COINS&Tools& MICIS/ DICOM/Receiver/ Assessment/Manager/ Self/Assessment/ Tablet/ Query/Builder/ Portals/ CAS/ My/Security/ Been/in/use/and/iteraFvely/improved/for/ over/7/years/ Fully/Open/Source/stack/(LAPP)/ Current/counts/of/important/arFfacts/ /PIs/82/ /Studies//334/(expired/and/onRgoing)/ /Subjects//~///15,000/ /Scan/Sessions/~//19,000/ /Clinical/Assessments/~/183,000/ 5/minutes/of/resFngRstate/ data/from/healthy/ subjects/combined/across/ studies/ Datasets/decomposed/in/a/ single/group/ica/ 28/components/(RSNs)/ selected/for/analysis/ Examined/how/baseline/ factors/such/as/age/and/ gender/affect/funcfonal/ connecfvity/ 2012 UMN A. Scott, W. Courtney, D. Wood, R. De la Garza, S. Lane, R. Wang, J. Roberts, J. A. Turner, and V. D. Calhoun, "COINS: An innovative informatics and neuroimaging tool suite built for large heterogeneous datasets," Frontiers in Neuroinformatics, vol. 5, pp. 1-15, UMN E. Allen, et al, "A baseline for the multivariate comparison of resting state networks," Frontiers in Systems Neuroscience, vol. 5:2, 2011.

11 FuncFonal/ConnecFvity/Features/ MulFvariate/TesFng/Framework/ Component/spaFal/maps/ FuncFonal/network/connecFvity/(FNC)/ univariate/models/ (1) H 0 ://β 1 /=/0 / :/ (v)/h 0 : // β v /=/0/ Is/voxel/i#affected#by/age?# mulfvariate/models/ H 0 ://β 1 /=/β 2 /=/ /β v #=/0/ / / Are/any/of/these/voxels/affected/by/age?/ Y&=&& Y&=&& Y&=&& Time/course/spectra/ 2012 UMN 2012 UMN Univariate/Followup/ examples/ Relationship to Disease (N=1140) Healthy (N=590) Substance Use (N=469) Schizo/BP (N=81) 2012 UMN E. Allen, et al, "A baseline for the multivariate comparison of resting state networks," Frontiers in Systems Neuroscience, vol. 5:2, 2011.

12 Within network example: anterior DMN Between network example: precuneus-cerebellar connectivity Spatial Sorting: Default Mode Mask Using wfu pickatlas to define mask using regions reported in Rachle 2001 paper Posterior parietal cortex BA7 Occipitoparietal junction BA 39 Precuneus Posterior cingulate Frontal Pole BA 10 Smooth in SPM with same kernel used on fmri data Sort in GIFT using spatial sorting ICA to identify Default Mode Network Healthy Healthy vs Schizo (N=26/26) Schizo +Symptoms A.Garrity, G.D.Pearlson, K.McKiernan, D.Lloyd, K.A.Kiehl, and V.D.Calhoun, "Aberrant A.Garrity, G.D.Pearlson, K.McKiernan, D.Lloyd, K.A.Kiehl, and V.D.Calhoun, "Aberrant 'Default Mode' Functional Connectivity in Schizophrenia," Am. J. Psychiatry, 'Default Mode' Functional Connectivity in Schizophrenia," Am. J. Psychiatry,

13 Extracting Diagnostic Information from Intrinsic Networks Robustness of modes.5 khz 1 khz tone, sweep, whistle Accurate classification requires singlesubject accuracy -> very stringent requirement! We cannot use knowledge of the diagnosis in the development of the 49 Standard Standard Target classification algorithm Modes Discriminate Between Schizophrenia and Bipolar Disorder," Hum. Brain Map., vol. 29, pp , Standard Standard Standard Novel Standard V. D. Calhoun, G. D. Pearlson, P. Maciejewski, and K. A. Kiehl, "Temporal Lobe and 'Default' Hemodynamic Brain Temporal Lobe Synchrony Supervised Classification Step 1: Select Training Group Step 2: Use ICA to extract temporal lobe maps Step 3: Compute within-group mean images Step 4: Subtract the mean images Step 5: Set a positive and negative threshold Temporal Lobe Synchrony in Schizophrenia Step 6: Form classification measure (average the values within each boundary and subtract) ( ) ( ) DF t+, t, i = µ i, IM > t + µ iim, < t.* MSK HC SZ Step 7: Optimize group discrimination (using a sensible error metric) ( + ) = ( ( + ) > ) + ( ( + ) < ) min Err t, t DF t, t, i 0 DF t, t, i 0 i Sz Step 8: Apply classification to new data i Hc HC 1 HC N t ICA t + Sz 1 Sz N Calhoun VD, Kiehl KA, Liddle PF, Pearlson GD: Aberrant Localization of Synchronous Hemodynamic Calhoun VD, Kiehl KA, Liddle PF, Pearlson GD: Aberrant Localization of Synchronous Hemodynamic Activity in Auditory Cortex Reliably Characterizes Schizophrenia. Biol Psychiatry 2004; Activity in Auditory Cortex Reliably Characterizes Schizophrenia. Biol Psychiatry 2004;

14 Temporal Sorting: fbirn SIRP Task Methods Subjects & Task 28 subjects (14 HC/14 SZ) across two sites Three runs of SIRP task preprocessed with SPM2 ICA Analysis All data entered into group ICA analysis in GIFT ICA time course and image reconstructed for each subject, session, and component Images: sessions averaged together creating single image for each subject and component Time courses: SPM SIRP model regressed against ICA time course Statistical Analysis: Images: 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 ANOVA Component 1: Bilateral Frontal/Parietal fbirn Phase II Data: fbirn Phase II Data: NCRR (NIH), 5 MOI RR ( ) and 1 U24 RR (2006 onwards) 53 NCRR (NIH), 5 MOI RR ( ) and 1 U24 RR (2006 onwards) 54 Component 2: Right Frontal, Left Parietal, Post. Cing. Component 3: Temporal Lobe fbirn Phase II Data: fbirn Phase II Data: NCRR (NIH), 5 MOI RR ( ) and 1 U24 RR (2006 onwards) 55 NCRR (NIH), 5 MOI RR ( ) and 1 U24 RR (2006 onwards) 56

15 Example 2: Simulated Driving Paradigm Previous Work * Drive Watch Walter, Driving Watching Our results suggest that simulated driving engages mainly areas concerned with perceptualmotor integration and does not engage areas associated with higher cognitive functions. 57 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. 58 Baseline Simulated Driving Results SPM 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. * 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 , 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

16 Error Monitoring & Inhibition motivation risk assessment internal space Exponential Decrease During Driving Low Order Visual High Order Visual visuomotor integration Increased during Driving Increase less during Watching Interpretation of Results Driving Speed EtOH Time Over Max. Speed Vigilance Visual Monitoring Increased during epoch transitions spatial attention, monitoring dorsal visual stream ventral visual stream external space Decreased Only During Driving Motor Control gross motor control motor planning Motor Coordination fine motor control visuomotor integration Increased only during Driving 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 Constrained ICA T 1 T λ W = ρ EG { ( ) } Eg { y Wx x µ y( y : W) x } 2 Q. Lin, J. Liu, Y. Zheng, H. Liang, and V. D. Calhoun, "Semi-blind Spatial ICA of fmri Using Spatial Constraints, Hum. Brain Map., vol. 31, Individual variability Current Directions Software Dependencies in space, time, space/time Dynamics Left Hemisphere Visual Stimuli Onset unique downloads Funded by: 1R01EB

17 DiagnosFc/ClassificaFon:/FNC/ How/informaFve/is/5/minutes/of/resFngRstate/fMRI/data?/ Training/data:/16/controls/+/16/SZ/paFents/ TesFng/data:/12/controls/+/12/paFents/ / Separate/ICAs/performed/on/training/tesFng/sets,/9/RSNs/selected/ / Features/are/the/temporal/correlaFons/between/components/ mean/correlafon/ trstafsfc/(controls/ /pafents)/ DiagnosFc/ClassificaFon:/FNC/ How/informaFve/is/5/minutes/of/resFngRstate/fMRI/data?/ Method/ Overall/Error/(%)/ SensiFvity/(%)/ Specificity/(%)/ Linear/Discriminant/ 29.17/ 41.67/ 100/ SVM/(linear)/ 16.67/ 75/ 91.67/ SVM/(RBF)/ 4.17/ 100/ 91.67/ K/nearest/neighbors/ 0/ 100/ 100/ Naïve/Bayes/ 20.83/ 75/ 83.33/ Neural/Net/(bp)/ 0/ 100/ 100/ Mohammad/R./Arbabshirani/and/Vince/D./Calhoun,/In/submission/ Mohammad/R./Arbabshirani/and/Vince/D./Calhoun,/In/submission/ 2012 UMN 2012 UMN

18 Methods for Identifying Resting Networks 2012 UMN E. Erhardt, E. Allen, E. Damaraju, and V. D. Calhoun, "On network derivation, classification, and visualization: a response to Habeck and Moeller," Brain Connectivity, vol. 1, pp. 1-19,