1 / 18

Multimodal Neuroimaging Training Program An fMRI study of visual search

Multimodal Neuroimaging Training Program An fMRI study of visual search Functional Magnetic Resonance Imaging: Group J. Wenzhu Bi, MS Graduate Student Biostatistics, CNBC University of Pittsburgh. Yanni Liu, PhD Graduate Student/Post-doc Psychology University of Michigan.

Download Presentation

Multimodal Neuroimaging Training Program An fMRI study of visual search

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Multimodal Neuroimaging Training Program An fMRI study of visual search Functional Magnetic Resonance Imaging: Group J Wenzhu Bi, MS Graduate Student Biostatistics, CNBC University of Pittsburgh Yanni Liu, PhD Graduate Student/Post-doc Psychology University of Michigan David Roalf, BS Graduate Student Behavioral Neuroscience Oregon Health Science Univ. Xingchen Wu, MD & PhD DRCMR, MR Dept. Copenhagen University Hospital Hvidovre Denmark

  2. Aims and Methods Aims -Learn to implement block and event-related fMRI experimental designs -Learn fMRI data pre-processing steps -Learn fMRI data post-processing: GLM and group analysis Methods -Subjects scanned: n=6 (3 males, 3 females) -Scanner: Siemens 3T -Images collected: MPRAGE(T1), In-Plane(T2 anatomical), EPI-BOLD(T2*,interleaved acquisition, TR=2s, voxel size 3.2mm3) - Block Design: 166 volumes X 4 runs - Event-Related Design: 159 volumes X 4 runs -Functional analysis: WashU pre-processing script, AFNI

  3. Task and Hypotheses -Visual Search attention task (feature vs. conjunction search) -More demanding attention task will elicit larger RT/Lower Accuracy -More demanding attention task result in greater activation of attention network (parietal regions) Is there anE? Conjunction Feature vs

  4. Treisman & Gelade 1980 Behavioral Results t(6)=3.63, p<.02 t(6)=2.74, p<.04

  5. Design F C Block ER Wager, 2007 4 runs X 6 blocks X 10 trials 4 runs X 4 same task sets X 12 trials • Pros: • High detection power due to response summation. • Simple analysis Con: • Can’t look at effects of single events (e.g., correct vs. incorrect trials; target present vs. absent) • Pros: • Good estimation of time courses and reasonable detection • Enables post hoc sorting (e.g., correct vs. incorrect; target present vs. absent) Con: • Some loss of power for the contrast between trial types.

  6. Pre/Post Processing • Post-processing • Individual analysis • GLM analysis • Assumed HRF model • Deconvolution (Finite Impulse Response) • ROI analysis • Group Analysis • Wilcoxon test • Pre-processing • Slice timing correction (Sinc interpolation) • Motion correction • Intensity scaling • Spatial smoothing • Spatial normalization (Talairach atlas transformation)

  7. Block Data Example Conjunction Feature Conj. vs Feat. L L L Feat. > baseline Conj. > baseline Conj. > Feat. Conj. < Feat. Conj. < baseline Feat. < baseline R R R q = 0.05

  8. Block vs. ER Data Block design ER design Results: Block design is more powerful to detect cerebral activation than ER design. ER design allows us to examine individual trial responses. L L Conj. > Feat. Conj. < Feat. R R q = 0.05 Conjunction HRF Feature HRF

  9. Spatial Smoothing A Gaussian filter with FWHM (full-width-half-max) = 6.4mm (i.e., twice the voxel width). Pros: -Smoothing resulted in greater areas of activation. -Increased signal to noise ratio Cons: -Reduced spatial precision -Introduce statistical interdependence among voxels R L Smoothed Conj. > Feat. Conj. < Feat. L R Non-smoothed FDR q=0.05

  10. Group Analysis: Block Design -Individual subject data was transformed to a standard space (Talairach). -A non-parametric Wilcoxon Signed Rank test was used to test for difference in visual search. L Wilcoxon Statistical map, |Z|>1.964, n=6 Conj. > Feat. L L Conj. < Feat. Non-Smoothed L L Smoothed

  11. Feature Conjunction ROI Timecourse Data Block onset Block offset n=6 Left Occipital Lobe (2096 mm3) TR n=6 TR Right Parietal Lobe (1263 mm3)

  12. What we have learned • We learned the details of fMRI pre-processing steps. This course allowed for discussion and understanding of slice-time correction, motion correction, spatial smoothing • We learned the details of post-processing including the use of the GLM for modeling our fMRI experiment. We also learned the analysis of individual and group level data. • AFNI- A good tool for understanding the complicated steps of analysis. • There is no recipe for fMRI analysis. Each study design and each analysis is unique which requires detailed understanding of the processing steps.

  13. Acknowledgements • Seong-Gi Kim • William Eddy • Mark E. Wheeler • Jeff Phillips • Elisabeth Ploran • Denise Davis • Tomika Cohen • Rebecca Clark

  14. How much movement is too much? Depends on many things: -the type of movement (sharp movement vs. drift) -timing of the movement (during a trial vs. during a break period) -the resolution of your data: 3 mm movement may be okay if you are collecting 3.2 X 3.2 X 3.2 mm3 resolution but may not if you are collecting 1.0 X 1.0 X 1.0 mm3 No specific criteria, the investigator must decide!!

  15. Deconvolution Assumed HRF

  16. Standardization Subject1 Subject 2 Subject 3

  17. Motor Analysis Left Hand Response Right Hand Response

More Related