Pushing spatial and temporal resolution for functional and diffusion MRI in the Human Connectome Project - PubMed (original) (raw)
doi: 10.1016/j.neuroimage.2013.05.012. Epub 2013 May 21.
Junqian Xu, Edward J Auerbach, Steen Moeller, An T Vu, Julio M Duarte-Carvajalino, Christophe Lenglet, Xiaoping Wu, Sebastian Schmitter, Pierre Francois Van de Moortele, John Strupp, Guillermo Sapiro, Federico De Martino, Dingxin Wang, Noam Harel, Michael Garwood, Liyong Chen, David A Feinberg, Stephen M Smith, Karla L Miller, Stamatios N Sotiropoulos, Saad Jbabdi, Jesper L R Andersson, Timothy E J Behrens, Matthew F Glasser, David C Van Essen, Essa Yacoub; WU-Minn HCP Consortium
Affiliations
- PMID: 23702417
- PMCID: PMC3740184
- DOI: 10.1016/j.neuroimage.2013.05.012
Pushing spatial and temporal resolution for functional and diffusion MRI in the Human Connectome Project
Kamil Uğurbil et al. Neuroimage. 2013.
Abstract
The Human Connectome Project (HCP) relies primarily on three complementary magnetic resonance (MR) methods. These are: 1) resting state functional MR imaging (rfMRI) which uses correlations in the temporal fluctuations in an fMRI time series to deduce 'functional connectivity'; 2) diffusion imaging (dMRI), which provides the input for tractography algorithms used for the reconstruction of the complex axonal fiber architecture; and 3) task based fMRI (tfMRI), which is employed to identify functional parcellation in the human brain in order to assist analyses of data obtained with the first two methods. We describe technical improvements and optimization of these methods as well as instrumental choices that impact speed of acquisition of fMRI and dMRI images at 3T, leading to whole brain coverage with 2 mm isotropic resolution in 0.7 s for fMRI, and 1.25 mm isotropic resolution dMRI data for tractography analysis with three-fold reduction in total dMRI data acquisition time. Ongoing technical developments and optimization for acquisition of similar data at 7 T magnetic field are also presented, targeting higher spatial resolution, enhanced specificity of functional imaging signals, mitigation of the inhomogeneous radio frequency (RF) fields, and reduced power deposition. Results demonstrate that overall, these approaches represent a significant advance in MR imaging of the human brain to investigate brain function and structure.
Copyright © 2013 Elsevier Inc. All rights reserved.
Figures
FIGURE 1
SNR achievable at 3 Tesla in a monopolar diffusion weighted spin-echo sequence at different maximal gradient strengths (Gmax), normalized to Gmax=100 mT/m. The monopolar dMRI sequence is schematically depicted in Figure 1. Ramp times shown in the pulse sequence diagram were ignored for the calculations (i.e. they were assumed to be infinitely fast). The minimum δ (see diagram) was calculated for a given b, G and d (where d is the separation between the two gradient pulses and Δ = δ+d, ignoring the ramp times) by solving 0 = b − (2_π_ · 42.58 ×10− ·G_· δ ) · 10− · (2_δ /3+ d) where b is s/mm2, δ and d in ms and G in mT/m, with d=6 ms. The minimum TE was calculated as 2δ+TE0, where TE0 is the minimum TE achievable with δ = 0; TE0 was taken to be 15 ms based on our existing sequence with partial Fourier acquisition). SNR is calculated using the biexponential diffusion approximation using the equation: SNR ∝ (0.75_e_−bD F + 0.25_e_−bD S)_e_− (δ + _TE_0)/_T_2 where DF_ and DS are fast and slow diffusion constants, respectively, (assumed to be 0.8×10−3 and 1×10−4 mm2/s) with corresponding fractional pool sizes of 0.75 and 0.25 (taken from (Ronen et al., 2005)),. White matter 3T T2 was assumed to be 70 ms (Stanisz et al., 2005).
FIGURE 2
Slices (7 Tesla, 0.8 mm isotropic resolution) from a single-shot EPI images (MB=1). 30 slices out of 128 are shown. Data presented from a single scan (without averaging) covering the whole brain using consecutively acquired single slices. Data obtained using Siemens SC72 “whole body” gradients. Acquisition parameters were: matrix size = 262×262, 128 slices, FOV = 210×210 mm2, TE/TR = 22ms/9350ms, in-plane phase encode acceleration (iPAT)=4, 5/8 partial Fourier, 96 ACS lines. Image supplied by An Vu from data obtained in CMRR.
FIGURE 3
Functional maps at 7 Tesla obtained with slice and phase-encode acceleration. Two representative coronal slices showing functional activation maps obtained with 16 fold two dimensional acceleration (4 fold slice (i.e. MB=4) and 4 fold in-plane phase encode accelerations) for a complex visuo-motor dissociation task; 90 slices were acquired in 1.5 sec with 1×1 mm2 in plane resolution 2 mm slice thickness. A total of 252 images were obtained with the subjects performing the task during three blocks of on-off periods. Maximal aliasing in these data was 16 fold. Adapted from (Moeller et al., 2008; Moeller et al., 2010).
FIGURE 4
Combining Multiband and SIR accelerations: Multiplex EPI images obtained at 3 Tesla. 4 adjacent slices are shown out of the total 60 slices obtained with 2 mm isotropic resolution covering the entire brain. Each row of images was obtained with different MB and SIR accelerations, producing simultaneous acquisition of 1, 4, 6 and 12 slices in a single EPI echo train. Adapted from (Feinberg et al., 2010).
FIGURE 5
Improvements in detection of RSNs (resting state networks) with shorter TRs. A single RSN is displayed at the same statistical threshold obtained at 3T from three 10 min acquisition periods obtained with different TRs using Multiplexed EPI. Data were gathered in a single session from one subject, using 3 mm isotropic resolution. TR=2.5 s (no acceleration), TR= 0.8 s (4 fold slice acceleration), and TR=0.4 s (9 fold slice acceleration). The color overlays are z-statistic images, thresholded at Z = 4 in all cases. Adapted from (Feinberg et al., 2010).
FIGURE 6
Controlled aliasing to displace simultaneously acquired images relative to each other to improve subsequent unaliasing. Multiband EPI (MB2) images acquired on the 3T WU-Minn consortium HCP scanner, without (top row) and with (bottom row) controlled aliasing (blipped CAIPI) using a PESHIFT of FOV/2. Slices from three different regions of the brain are shown. Images supplied by Junqian Xu from data acquired in CMRR.
FIGURE 7
Slice accelerated Multiband images at 3T at different acceleration factors. Three slices from a 2 mm isotropic resolution, 64 slice whole brain data set obtained with slice acceleration up to MB factor of 12. For comparison, images were acquired with the same TR (4.8 s) based on the minimum TR attainable with standard EPI (i.e. MB=1). The example axial slices shown were not from the same MB slice group. Achievable TR at a given MB factor is listed below the MB factors given to show the acceleration potential. Adapted from (Xu et al., 2012a, b).
FIGURE 8
Comparing 6-fold slice accelerated Multiband images at 3T with unaccelerated standard acquisition. Selected slices from a 1.6 mm isotropic, 80 slice whole brain data set obtained with Multiband EPI with PESHIFT=FOV/3, MB factor 6 and standard EPI (MB=1). TE=30 ms; 6/8 Partial Fourier along phase encode direction. TR =6.7 s for both, set by the minimum TR attainable with MB=1. Minimum TR that would be possible with MB=6 acquisition with these parameters would be 1.1 s. Data was obtained with a 32 channel coil on the 3T WU-Minn HCP scanner. Adapted from (Xu et al., 2012a, b).
FIGURE 9
Noise amplifications due to unaliasing of slice accelerated Multiband EPI data. Noise amplification is calculated as a “g-factor” and presented as a histogram. MB=2 (green), MB =4 (black), MB=8 (blue), and MB=12 (red). Data obtained with a 32 channel coil on the 3T WU-Minn HCP scanner. Adapted from (Xu et al., 2012a, b).
FIGURE 10
Quantifying residual aliasing among simultaneously acquired slices. Signal leakage (_L_-factor) maps showing residual aliasing among simultaneously acquired slices at 3T for MB3, MB4, MB8 and MB12 with with PESHIFT.The oscillation imposed on slice (appears in red/yellow color) “leaks” into other simultaneously acquired slices due to resdiual aliasing. Adapted from (Moeller et al., 2012).
FIGURE 11
Effect of calibration (i.e. reference) scan on image reconstruction of slice accelerated multiband images at 7T. A coronal cross sections of axially acquired 7T whole brain data set acquired with slice and in-plane-phase encode accelerated Multiband EPI using different calibration scans: Top row: GRE (i.e. FLASH) calibration scan, Lower row: EPI calibration scans. Data were acquired in the same session from the same subject. Each volumetric image set had its own calibration scan. Multiband EPI was obtained with MB=3, in plane acceleration =3, TE/TR=20ms/2740ms, 1.1 isotropic resolution, 8/6 partial Fourier along phase encode direction, 123 slices. The GRE (FLASH) calibration scans had “matched” TE/TR 20ms/2740ms, 1.1×2.4×1.1mm resolution, and FOV of 209×209×135mm as in EPI.For the EPI the single band ACS data was acquired with 48 ACS lines.
FIGURE 12
Temporal stability in dMRI data acquired at 3T with slice and in-plane acceleration. A single slice as a function of time is shown from a monopolar dMRI acquisition with a single b value and direction (b=1500 s/mm2); MB=3 and in-plane phase encode acceleration of 3, 2 mm isotropic resolution, repeated consecutively in time. Acquired on the 3T WU-Minn HCP scanner. Adapted from (Moeller et al., 2013).
FIGURE 13
Reduction of peak power using Time-Shifted Multiband pulses: (A) Pulse shapes, from left to right: conventional four-banded (MB4) pulse of duration t, composed of four single-banded (SB) sinc RF pulse (R = 5.2) with band #1 (no frequency offset) and band #4 ~3 kHz frequency offset; the same conventional MB4 pulse with stretched duration of 1.75·t, time-shifted MB4 pulse generated for duration 1.75·t (25% temporal shift between bands). For all plots, solid lines represent the magnitude, dashed lines the real component, and dotted lines the imaginary component. (B) Plot of required peak B1 vs. total pulse duration for four-banded pulses with 3 kHz inter-band frequency offsets: stretched conventional pulse (dark blue), time-shifted pulse (green), time-shifted with static (i.e. fixed) inter-band phase offsets (light blue), and time-shifted with optimized phase offsets for each shift (red). B1 and duration are shown relative to the base single-banded sinc pulse. Note that time shifted pulses ultimately achieve the same peak B1 whether or not phase optimization is employed. Adapted from (Auerbach et al., 2013).
FIGURE 14
Overcoming peak power limitation in Multiband spin-echo acquisition using Time-Shifted Pulses. Comparison of human brain images acquired with multi-band slice accelerations of 3–6 (MB3-MB6) using conventional multi-banded RF pulses (left column) and time-shifted RF pulses (right column) with equivalent effective bandwidth. The display window and level are the same for all images. Adapted from (Auerbach et al., 2013).
FIGURE 15
7 Tesla transmit B1 maps in the human head. The maps were generated by a 4 port driven TEM coil similar to that described in (Vaughan et al., 2001), except in this case the internal diameter of the coil was slightly larger to accommodate a 16 channel receive-only array within it. The transmitter generates a homogenous B1 when it is not loaded with the human head.
FIGURE 16
Using parallel transmit to simultaneously improve transmit B1 and constrain total RF energy of multiband pulses. (A) L curves quantifying tradeoffs between total RF power and excitation errors (i.e., root mean square error (RMSE)) in the human brain at 7T for Full pTx MB (○) and MB B1 shim (×), along with Circularly polarized (CP) mode (□) located at the crossing between the horizontal and vertical lines, labeled “Same fidelity” and “ Same Power”, respectively. Calculations were performed for a 7T coil described previously (Adriany et al., 2008) with azimuthally distributed 16 transmit and receive microstrip elements (B) MB2 RF excitation in the human brain at 7T for CP mode conventional MB, MB B1 shim and Full pTx MB pulses with the same total RF power. The pulses were designed for the coil described in (A) using subject dependent B1 calibration maps. The pulses were based on a single spoke design, with the base pulse being a filtered SINC pulse of BWTP = 10. Pulse duration = 1 ms, nominal flip angle = 6°, and inter-slice distance = 56 mm.
FIGURE 17
Default Mode Network (DMN) extracted from 1 mm isotropic resolution 7T resting-state fMRI time series. DMN, shown on three orthogonal slices, was extracted by ICA from 7T resting state data acquired using standard EPI without slice acceleration but with 4-fold acceleration along the in-plane phase encode direction. The DMN data in color is superimposed on 0.6 mm isotropic T1 weighted anatomical images obtained with MPRAGE at 7T. Adapted from (De Martino et al., 2011).
FIGURE 18
7 Tesla Multiband EPI data acquired with concurrent slice and in-plane phase encode acceleration. Three orthogonal slices are shown from a 1 mm isotropic resolution whole brain 7 Tesla Multiband EPI data obtained with a 32 channel receive array, MB=4, in-plane phase encode acceleration of 3, PESHIFT of FOV/3.
FIGURE 19
Detection of an RSN at 3T vs. 7T on the same subject. Three orthogonal slices extracted from 7 and 3 Tesla Multiband rfMRI on the same subject depicting the same RSN. The data were processed without any spatial smoothing. The 3T parameters were: 2mm isotropic, TR=1.37s, 1000 time-points, ICA auto-dimensionality of d=200; 7 T parameters:1.25mm, TR=2s, 450 time-points, ICA auto-dimensionality of d=81.
FIGURE 20
Diffusion weighted imaging at 7T vs. 3T; calculation of expected SNR as a function of echo time. Ratio of 7T SNR relative to 3T SNR for the central k-space point, ko, calculated for a spin echo sequence (90 deg. excitation and a single 180 deg. refocusing pulse) as a function of tek1, the time to the first k-space point. 6/8 Partial Fourier EPI acquisition was modeled. The echo time TE is equal to tek1 plus the time to acquire 2/8 k-space points to reach the ko signal where maximum echo amplitude occurs. Note that for the above graph, TE>0 even when tek1=0 which accounts for the less than approximately linear gain in SNR at tek1=0. T1 of white matter was taken as 0.838 s at 3T (Wansapura et al., 1999) and 1.22 s at 7T (Rooney et al., 2007). Bandwidth was kept the same. The minimum TE was taken as 15 ms. The T2 at 3T was reported to be between 55 and 69 ms ((Stanisz et al., 2005) and references therein); for these calculations, it was assumed to be the average of this range, i.e. 62 ms. The 7T white matter T2 was taken as 45 ms (Yacoub et al., 2003). Intrinsic SNR was scaled 1.15 times the magnetic field as previously shown experimentally and by modeling (Vaughan et al., 2001). Same matrix size at 3T and 7T was assumed.
FIGURE 21
Compressed sensing in q-space; tractography with half the acquired q-space data. Example of whole brain deterministic tractography using the Tensorline algorithm (Lazar et al., 2003) as implemented in TrackVis (
) from a complete (Left) 7T dMRI dataset (voxel size 1.5 mm isotropic, 128 diffusion gradients, b-value =1500s/mm2 and 15 b0 values), and the same dataset reconstructed from half of the volumes (64 diffusion gradients) and a compressed sensing acceleration factor R = 2 (Right). Very few differences can be observed.
FIGURE 22
Compressed sensing in q-space; effect on fiber orientation estimation. Uncertainty associated with the estimation of each probabilistic fiber orientation (3 fibers were used). The top row corresponds to the full dMRI datasets (R=1). The bottom row corresponds to the dataset reconstructed from half of the original dataset (R=2). Light blue areas correspond to an error close to 0° while the red areas correspond to an error close to 90°. Uncertainty is very similar even when using only half of the original dMRI dataset (bottom row), when compared to the full dataset (top row).
Similar articles
- Resting-state fMRI in the Human Connectome Project.
Smith SM, Beckmann CF, Andersson J, Auerbach EJ, Bijsterbosch J, Douaud G, Duff E, Feinberg DA, Griffanti L, Harms MP, Kelly M, Laumann T, Miller KL, Moeller S, Petersen S, Power J, Salimi-Khorshidi G, Snyder AZ, Vu AT, Woolrich MW, Xu J, Yacoub E, Uğurbil K, Van Essen DC, Glasser MF; WU-Minn HCP Consortium. Smith SM, et al. Neuroimage. 2013 Oct 15;80:144-68. doi: 10.1016/j.neuroimage.2013.05.039. Epub 2013 May 20. Neuroimage. 2013. PMID: 23702415 Free PMC article. - Pushing the limits of in vivo diffusion MRI for the Human Connectome Project.
Setsompop K, Kimmlingen R, Eberlein E, Witzel T, Cohen-Adad J, McNab JA, Keil B, Tisdall MD, Hoecht P, Dietz P, Cauley SF, Tountcheva V, Matschl V, Lenz VH, Heberlein K, Potthast A, Thein H, Van Horn J, Toga A, Schmitt F, Lehne D, Rosen BR, Wedeen V, Wald LL. Setsompop K, et al. Neuroimage. 2013 Oct 15;80:220-33. doi: 10.1016/j.neuroimage.2013.05.078. Epub 2013 May 24. Neuroimage. 2013. PMID: 23707579 Free PMC article. - The Lifespan Human Connectome Project in Aging: An overview.
Bookheimer SY, Salat DH, Terpstra M, Ances BM, Barch DM, Buckner RL, Burgess GC, Curtiss SW, Diaz-Santos M, Elam JS, Fischl B, Greve DN, Hagy HA, Harms MP, Hatch OM, Hedden T, Hodge C, Japardi KC, Kuhn TP, Ly TK, Smith SM, Somerville LH, Uğurbil K, van der Kouwe A, Van Essen D, Woods RP, Yacoub E. Bookheimer SY, et al. Neuroimage. 2019 Jan 15;185:335-348. doi: 10.1016/j.neuroimage.2018.10.009. Epub 2018 Oct 15. Neuroimage. 2019. PMID: 30332613 Free PMC article. - The parcellation-based connectome: limitations and extensions.
de Reus MA, van den Heuvel MP. de Reus MA, et al. Neuroimage. 2013 Oct 15;80:397-404. doi: 10.1016/j.neuroimage.2013.03.053. Epub 2013 Apr 1. Neuroimage. 2013. PMID: 23558097 Review. - Learning and comparing functional connectomes across subjects.
Varoquaux G, Craddock RC. Varoquaux G, et al. Neuroimage. 2013 Oct 15;80:405-15. doi: 10.1016/j.neuroimage.2013.04.007. Epub 2013 Apr 11. Neuroimage. 2013. PMID: 23583357 Review.
Cited by
- Heterogenous brain activations across individuals localize to a common network.
Peng S, Cui Z, Zhong S, Zhang Y, Cohen AL, Fox MD, Gong G. Peng S, et al. Commun Biol. 2024 Oct 5;7(1):1270. doi: 10.1038/s42003-024-06969-x. Commun Biol. 2024. PMID: 39369118 Free PMC article. - The degenerate coding of psychometric profiles through functional connectivity archetypes.
Di Plinio S, Northoff G, Ebisch S. Di Plinio S, et al. Front Hum Neurosci. 2024 Sep 10;18:1455776. doi: 10.3389/fnhum.2024.1455776. eCollection 2024. Front Hum Neurosci. 2024. PMID: 39318702 Free PMC article. - qPRF: A system to accelerate population receptive field decoding.
Waz S, Wang Y, Lu ZL. Waz S, et al. bioRxiv [Preprint]. 2024 Aug 15:2024.08.13.607805. doi: 10.1101/2024.08.13.607805. bioRxiv. 2024. PMID: 39185219 Free PMC article. Preprint. - Main Paths of Brain Fibers in Diffusion Images Mixed with a Noise to Improve Performance of Tractography Algorithm-Evaluation in Phantom.
Shirazinodeh A, Faraji H, Sharifzadeh Javidi S, Jafari AH, Nazemzadeh M, Saligheh Rad H. Shirazinodeh A, et al. J Biomed Phys Eng. 2024 Aug 1;14(4):357-364. doi: 10.31661/jbpe.v0i0.2108-1387. eCollection 2024 Aug. J Biomed Phys Eng. 2024. PMID: 39175552 Free PMC article. - Contributions of network structure, chemoarchitecture and diagnostic categories to transitions between cognitive topographies.
Luppi AI, Singleton SP, Hansen JY, Jamison KW, Bzdok D, Kuceyeski A, Betzel RF, Misic B. Luppi AI, et al. Nat Biomed Eng. 2024 Sep;8(9):1142-1161. doi: 10.1038/s41551-024-01242-2. Epub 2024 Aug 5. Nat Biomed Eng. 2024. PMID: 39103509 Free PMC article.
References
- Adriany G, Van de Moortele PF, Ritter J, Moeller S, Auerbach EJ, Akgun C, Snyder CJ, Vaughan T, Ugurbil K. A geometrically adjustable 16-channel transmit/receive transmission line array for improved RF efficiency and parallel imaging performance at 7 Tesla. Magn Reson Med. 2008;59:590–597. - PubMed
- Adriany G, Van de Moortele PF, Wiesinger F, Moeller S, Strupp JP, Andersen P, Snyder C, Zhang X, Chen W, Pruessmann KP, Boesiger P, Vaughan T, Ugurbil K. Transmit and receive transmission line arrays for 7 Tesla parallel imaging. Magn Reson Med. 2005;53:434–445. - PubMed
Publication types
MeSH terms
Grants and funding
- G0800578/MRC_/Medical Research Council/United Kingdom
- 098369/WT_/Wellcome Trust/United Kingdom
- P30 NS076408/NS/NINDS NIH HHS/United States
- 1U54MH091657/MH/NIMH NIH HHS/United States
- P41 EB015894/EB/NIBIB NIH HHS/United States
- F30 MH097312/MH/NIMH NIH HHS/United States
- U54 MH091657/MH/NIMH NIH HHS/United States
- P30 NS057091/NS/NINDS NIH HHS/United States
LinkOut - more resources
Full Text Sources
Other Literature Sources
Miscellaneous