1 / 98

Fluctuations and Networks: Thinking and Breathing

Fluctuations and Networks: Thinking and Breathing. Peter A. Bandettini, Ph.D. Section on Functional Imaging Methods & Functional MRI Facility National Institutes of Mental Health, NIH. Rasmus Birn. Overview. “Functional connectivity” Correlations in fMRI signal at rest

orlando-goodman
Download Presentation

Fluctuations and Networks: Thinking and Breathing

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. Fluctuations and Networks: Thinking and Breathing Peter A. Bandettini, Ph.D. Section on Functional Imaging Methods & Functional MRI Facility National Institutes of Mental Health, NIH

  2. Rasmus Birn Overview • “Functional connectivity” • Correlations in fMRI signal at rest • “Default Mode Network” • Respiration changes in fMRI • Respiration + Functional Connectivity • Modeling respiration-induced fMRI changes

  3. task t t The fMRI Signal Task-related Neuronal Activity Hemodynamics Spontaneous Neuronal Activity Thermal noise, Artifacts Physiologic noise (cardiac, respiration)

  4. Resting-state functional connectivity B. Biswal et al., MRM, 34:537 (1995) Activation during finger-tapping Correlations with “seed voxel” in motor cortex at rest Further work: M.J. Lowe, et al., NeuroImage 7(2), 1998. D. Cordes, et al., AJNR 21(9), 2000.

  5. “Default-Mode Network” PNAS 98, 2001

  6. Resting-state functional connectivity PNAS 100, 2003

  7. Deactivations Resting correlations • Are there differences? • Deactivation (blocked design) • Rest (functional connectivity) • Are deactivations related to the task, or to the brain “state”? • Mixed blocked / event-related Jason Diamond

  8. 2. Lexical Decision: “word” or “non-word”? Methods R.M. Birn, J.B. Diamond, M.A. Smith, P.A. Bandettini, NeuroImage 31, 2006 Task 1. Rest (eyes-closed) + FOIL 15 words/non-words (30s) FLURP LIST Fixation (30s) … + Blocked design (30s task, 30s rest)

  9. Results – Lexical Task Activations and De-activations during lexical task 20 Z -20

  10. Methods • Filter (respiration (0.3Hz), cardiac (1 Hz)) • Define ROI (e.g. deactivations in posterior cingulate) • Average time courses (at rest) in ROI • Correlate average time course with all voxels Functional Connectivity Analysis Lexical task Rest

  11. task t t The fMRI Signal Task-related Neuronal Activity Hemodynamics Spontaneous Neuronal Activity Thermal noise, Artifacts Physiologic noise (cardiac, respiration)

  12. 0 40 80 120 160 200 240 280 320 time (s) 400 B0 200 o DF =90 00 TIME = 110 s Field Map Figure courtesy of J. Bodurka Physiological fluctuations Cardiac M.S. Dagli et al., NeuroImage 9, 1999 Respiration time

  13. Correction of physiological noise RETROICOR (G. Glover et al., Magn. Reson. Med. 44, 2000.) Additional Regressors: MR Images sin( fc ) fc cos( fc ) Pulse Oximeter sin( 2fc ) cos( 2fc ) fr’ sin( fr ) Respiration cos( fr ) sin( 2fr ) 0 1 2 3 4 5 6 7 8 9 10 time (s) cos( 2fr )

  14. Methods • Filter (respiration (0.3Hz), cardiac (1 Hz)) • Define ROI (e.g. deactivations in posterior cingulate) • Average time courses (at rest) in ROI • Correlate average time course with all voxels Functional Connectivity Analysis Lexical task Rest

  15. 20 Z -20 Functional Connectivity Analysis 1 subject Activations during lexical task Correlation (of PC) at Rest 15 Z -15

  16. Functional Connectivity Analysis Group (n=10) Activations during lexical task Correlation (of PC) at Rest 10 6 Z |Z| 0 -10

  17. Overview • “Functional connectivity” • Correlations in fMRI signal at rest • “Default Mode Network” • Respiration changes in fMRI • Respiration + Functional Connectivity • Modeling respiration-induced fMRI changes

  18. 5 Z 0 5 % Breath-holding Group Maps (N = 7) MR Signal Respiration Cue time (s) 0 50 100 150 200 250 300 350 Breath-hold response (average Z-score) Anatomy

  19. 90% 0% Breath-hold vs. bolus contrast (Gd-DTPA) 20 2% MR Sig. 0 Signal(%) Resp -20 Cue -40 0 50 100 150 200 250 0 50 100 150 200 250 300 time (s) 40s 64s time(s) 18% 0%

  20. 10 -10 Breath-hold vs. bolus contrast (Gd-DTPA) Relative difference: DS (Gd-DTPA) – DS (Breath-hold) Arteries vs.Veins ?

  21. Resting fluctuations in respiration NeuroImage 21, 2004

  22. 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350 time (s) T max min 300 310 315 305 time (s) max - min RVT = T Estimating respiration volume changes Respiration time (s) Respiration Volume / Time (RVT)

  23. Resting fluctuations in respiration Correlation of fMRI time course with: RVT etCO2 BH

  24. +10% -10% Resting fluctuations in respiration Amplitude of BOLD signal correlated w/ RVT Z-score of BOLD signal correlated w/ RVT 6 |Z| 0 1 subject group (n=11) RVT = Respiration Volume per Time

  25. Power spectrum of RVT 0 0.05 0.1 0.15 0.2 0.25 Power spectrum of BOLD 0 0.05 0.1 0.15 0.2 0.25 frequency (Hz) Respiration changes co-localize Deactivations Resting-state corr. from seed ROI Respiration changes – corr. w/ RVT

  26. 6 10 10 Z |Z| Z Power spectrum of RVT 0 -10 0 0.05 0.1 0.15 0.2 0.25 Power spectrum of BOLD -10 0 0.05 0.1 0.15 0.2 0.25 frequency (Hz) Respiration changes co-localize Deactivations Resting-state corr. from seed ROI Respiration changes – corr. w/ RVT (1 subject) (1 subject)

  27. Correcting for changes in respiration • Regress out RVT • Keep respirations constant

  28. 0 0.1 0.2 Respiration (Rest) Power 0 0.2 0.4 0.6 0.8 1 Amplitude Frequency (Hz) 0 50 100 150 200 250 300 350 Power time (s) Frequency (Hz) fMRI Signal Power Signal 0 0.1 0.2 0 0.1 0.2 0 50 100 150 200 250 300 350 time (s) Frequency (Hz) fMRI Signal (after RVTcor) Power Signal 0 50 100 150 200 250 300 350 time (s) Frequency (Hz) Regress out Cardiac, Respiration, RVT

  29. 0 0.1 0.2 Respiration (Rest) Power 0 0.2 0.4 0.6 0.8 1 Amplitude Frequency (Hz) 0 50 100 150 200 250 300 350 Power time (s) Frequency (Hz) fMRI Signal Power Signal 0 0.1 0.2 0 50 100 150 200 250 300 350 time (s) Frequency (Hz) fMRI Signal (Constant Resp.) Signal Power 0 50 100 150 200 250 300 350 0 0.1 0.2 time (s) Frequency (Hz) Cue subject to keep breathing constant

  30. Standard Deviation over time Constant Resp + RETROICOR Constant Resp RETROICOR No correction RVTcor 5% 0% No corr - const 2% Difference Maps 0% -2%

  31. No correction RETROICOR RVTcor Constant Resp – no corr. Constant Resp - RETROICOR Standard Deviation over time Average Std. Dev. (averaged over all brain voxels, 10 subj) 2.5 2 Std. Dev. (%) 1.5 1 0.5

  32. No correction RETROICOR RVTcor 13 9 8 0 7 6 5 -13 -5 -6 Difference Maps -7 No correction RETROICOR -8 RVTcor -9 5 0 -5 Improving the detection of function Average Z-score (averaged over all active voxels, 10 subj) Z-score activations deactivations Z-score

  33. More Corrections? • RETROICOR (cardiac, respiration) • RVTcor (respiration volume / etCO2) • Motion parameters • Global detrending • White-matter detrending

  34. 10 Z 3.8 (p=10-4) Card+Resp Card+Resp Resp Vol Resp Vol Motion Motion Global Signal Global Signal WM Signal WM Signal Multiple physiological corrections Correlation with each regressor (n = 10) Differences in Std. Dev. when each regressor is removed +1% -0.1% -1%

  35. 12.00% 8.00% 4.00% WM detrend Resp Vol. / T. RETROICOR Motion detrend Global detrend 0.00% Resp.Vol./T. WM detrend RETROICOR Global detrend Motion detrend 12.00% 8.00% 4.00% 0.00% Relative contributions to noise Averaged over Gray Matter (4 subjects) Low Resolution (7.5 x 7.5 x 5 mm, 700 time points) High Resolution (1.7 x 1.7 x 4 mm, 80 time points)

  36. How much is the noise reduced? TSNR Std. Dev. over time 150 3% 0 0% Before Correction After Correction Before Correction After Correction +75 +3% +0.5% - 0.5% 0 - 3% Difference Difference

  37. Before correction After correction Improvement in temporal SNR (TSNR) TSNR 150 Low resolution (7.5 x 7.5 x 5 mm) High resolution (1.7 x 1.7 x 4 mm) Med. resolution (3.8 x 3.8 x 5 mm) 0 Before Correction After Correction 200 150 TSNR 100 +75 50 0 0 Difference

  38. How Improved TSNR Translates Murphy K, Bodurka J, Bandettini P, “How long to scan: The relationship between temporal signal to noise ratio and necessary scan duration,” NeuroImage (in press)

  39. SNR vs TSNR Bodurka J, Bandettini P, “Mapping the MRI voxel volume in which thermal noise matches physiological noise - implications for fMRI”, NeuroImage (in press)

  40. 15 Z -15 Resting-state correlations Areas correlated with posterior cingulate 1 subject after RETROICOR after RETROICOR + RVTcor 15 Z -15

  41. Resting-state correlations Areas correlated with posterior cingulate 1 subject Constant Respirations Remove global signal changes 15 15 Z Z -15 -15

  42. 6 10 Z |Z| 0 -10 BOLD signal changes – Group data Group data (n=10) Lexical Task Respiration changes

  43. Resting-state correlations Areas correlated with posterior cingulate Group data (n=10) after RETROICOR after RETROICOR + RVTcor 10 10 Z Z -10 -10

  44. Resting-state correlations Areas correlated with posterior cingulate Group data (n=10) Constant Respirations Remove global signal changes 10 10 Z Z -10 -10 Click here for unthresholded maps

  45. Resting-state correlations Areas correlated with posterior cingulate Group data (n=10) Constant Respirations Cued Variable Respirations 10 10 Z Z -10 -10 End

  46. A closer look at respiration changes • Breath-hold vs. Resting variations in breathing • The respiration “impulse response function” • How do respiration changes really affect the BOLD signal? (shape, latency, …) • Echo-time (TE) dependence

  47. 0 BOLD Z Respiration Cue 0 100 200 300 Time (sec) -4 5 Z BOLD Respiration RVT 0 0 100 200 300 Time (sec) Respiration induced signal changes Rest Breath-holding (N=7)

  48. 30 CC = 0.76 Rest (%) 10 0 0 20 10 Breath hold (%) Resting changes in breathing vs. Breath-holding Correlation with Respiration Volume / Time (RVT)

  49. 20 20s 20s 10 %S (Rest) 0 0 10 20 30 0s 0s %S (Breath-hold) 0 20 avg %S (Rest) -10 Latency: Rest (s) -3 -2 -1 0 0 0 10 20 30 Latency: Breath-hold (s) avg %S (Breath-hold) Resting changes in breathing vs. Breath-holding Rest vs. BH across subjects (averaged over space) Rest vs. BH within subject Latency Rest CC= 0.866 CC= 0.726 Latency Breath-hold CC= -0.446 CC= 0.618 Latency: Rest (s) Latency: Breath-hold (s)

  50. 0 50 100 150 200 250 300 350 time (s) 0 50 100 150 200 250 300 350 time (s) Respiration Changes vs. BOLD How are the BOLD changes to respiration variations? RVT ? fMRI Signal

More Related