Predicting the fMRI Signal Fluctuation with Recurrent Neural Networks Trained on Vascular Network Dynamics

被引:6
作者
Sobczak, Filip [1 ,2 ]
He, Yi [1 ,3 ]
Sejnowski, Terrence J. [4 ,5 ]
Yu, Xin [1 ,6 ,7 ]
机构
[1] Max Planck Inst Biol Cybernet, High Field Magnet Resonance Dept, Translat Neuroimaging & Neural Control Grp, D-72076 Tubingen, Germany
[2] Univ Tubingen, Grad Training Ctr Neurosci, Int Max Planck Res Sch, D-72074 Tubingen, Germany
[3] Danish Res Ctr Magnet Resonance, DK-2650 Hvidovre, Denmark
[4] Howard Hughes Med Inst, Salk Inst Biol Studies, Computat Neurobiol Lab, La Jolla, CA 92037 USA
[5] Univ Calif San Diego, Div Biol Sci, La Jolla, CA 92093 USA
[6] Massachusetts Gen Hosp, Athinoula A Martinos Ctr Biomed Imaging, Charlestown, MA 02129 USA
[7] Harvard Med Sch, Charlestown, MA 02129 USA
关键词
default mode network; machine learning; resting state; single-vessel; slow oscillation; RESTING-STATE FMRI; INDEPENDENT COMPONENT ANALYSIS; LOW-FREQUENCY FLUCTUATIONS; FALSE DISCOVERY RATE; HUMAN VISUAL-CORTEX; FUNCTIONAL CONNECTIVITY; GLOBAL SIGNAL; DEFAULT-MODE; BRAIN ACTIVITY; CEREBRAL-CORTEX;
D O I
10.1093/cercor/bhaa260
中图分类号
Q189 [神经科学];
学科分类号
071006 ;
摘要
Resting-state functional MRI (rs-fMRI) studies have revealed specific low-frequency hemodynamic signal fluctuations (<0.1 Hz) in the brain, which could be related to neuronal oscillations through the neurovascular coupling mechanism. Given the vascular origin of the fMRI signal, it remains challenging to separate the neural correlates of global rs-fMRI signal fluctuations from other confounding sources. However, the slow-oscillation detected from individual vessels by single-vessel fMRI presents strong correlation to neural oscillations. Here, we use recurrent neural networks (RNNs) to predict the future temporal evolution of the rs-fMRI slow oscillation from both rodent and human brains. The RNNs trained with vessel-specific rs-fMRI signals encode the unique brain oscillatory dynamic feature, presenting more effective prediction than the conventional autoregressive model. This RNN-based predictive modeling of rs-fMRI datasets from the Human Connectome Project (HCP) reveals brain state-specific characteristics, demonstrating an inverse relationship between the global rs-fMRI signal fluctuation with the internal default-mode network (DMN) correlation. The RNN prediction method presents a unique data-driven encoding scheme to specify potential brain state differences based on the global fMRI signal fluctuation, but not solely dependent on the global variance.
引用
收藏
页码:826 / 844
页数:19
相关论文
共 154 条
  • [61] Functional connectivity between task-positive and task-negative brain areas and its relation to working memory performance
    Hampson, Michelle
    Driesen, Naomi
    Roth, Jennifer K.
    Gore, John C.
    Constable, R. Todd
    [J]. MAGNETIC RESONANCE IMAGING, 2010, 28 (08) : 1051 - 1057
  • [62] Brain connectivity related to working memory performance
    Hampson, Michelle
    Driesen, Naomi R.
    Skudlarski, Pawel
    Gore, John C.
    Constable, R. Todd
    [J]. JOURNAL OF NEUROSCIENCE, 2006, 26 (51) : 13338 - 13343
  • [63] Periodic changes in fMRI connectivity
    Handwerker, Daniel A.
    Roopchansingh, Vinai
    Gonzalez-Castillo, Javier
    Bandettini, Peter A.
    [J]. NEUROIMAGE, 2012, 63 (03) : 1712 - 1719
  • [64] Functional connectivity dynamics: Modeling the switching behavior of the resting state
    Hansen, Enrique C. A.
    Battaglia, Demian
    Spiegler, Andreas
    Deco, Gustavo
    Jirsa, Viktor K.
    [J]. NEUROIMAGE, 2015, 105 : 525 - 535
  • [65] Electrophysiological correlates of the brain's intrinsic large-scale functional architecture
    He, Biyu J.
    Snyder, Abraham Z.
    Zempel, John M.
    Smyth, Matthew D.
    Raichle, Marcus E.
    [J]. PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2008, 105 (41) : 16039 - 16044
  • [66] The fMRI signal, slow cortical potential and consciousness
    He, Biyu J.
    Raichle, Marcus E.
    [J]. TRENDS IN COGNITIVE SCIENCES, 2009, 13 (07) : 302 - 309
  • [67] Ultra-Slow Single-Vessel BOLD and CBV-Based fMRI Spatiotemporal Dynamics and Their Correlation with Neuronal Intracellular Calcium Signals
    He, Yi
    Wang, Maosen
    Chen, Xuming
    Pohmann, Rolf
    Polimeni, Jonathan R.
    Scheffler, Klaus
    Rosen, Bruce R.
    Kleinfeld, David
    Yu, Xin
    [J]. NEURON, 2018, 97 (04) : 925 - +
  • [68] Spatio-Temporal Dynamics of Intrinsic Networks in Functional Magnetic Imaging Data Using Recurrent Neural Networks
    Hjelm, R. Devon
    Damaraju, Eswar
    Cho, Kyunghyun
    Laufs, Helmut
    Plis, Sergey M.
    Calhoun, Vince D.
    [J]. FRONTIERS IN NEUROSCIENCE, 2018, 12
  • [69] Low frequency BOLD fluctuations during resting wakefulness and light sleep: A simultaneous EEG-fMRI study
    Horovitz, Silvina G.
    Fukunaga, Masaki
    de Zwart, Jacco A.
    van Gelderen, Peter
    Fulton, Susan C.
    Balkin, Thomas J.
    Duyn, Jeff H.
    [J]. HUMAN BRAIN MAPPING, 2008, 29 (06) : 671 - 682
  • [70] Dynamic functional connectivity: Promise, issues, and interpretations
    Hutchison, R. Matthew
    Womelsdorf, Thilo
    Allen, Elena A.
    Bandettini, Peter A.
    Calhoun, Vince D.
    Corbetta, Maurizio
    Della Penna, Stefania
    Duyn, Jeff H.
    Glover, Gary H.
    Gonzalez-Castillo, Javier
    Handwerker, Daniel A.
    Keilholz, Sheila
    Kiviniemi, Vesa
    Leopold, David A.
    de Pasquale, Francesco
    Sporns, Olaf
    Walter, Martin
    Chang, Catie
    [J]. NEUROIMAGE, 2013, 80 : 360 - 378