Final Report Summary - R&B BRAIN (In-depth investigation of the non-stationary dynamics of resting-state brain activity and its impact on behaviour and cognition)
The emphasis of the current research fellowship is to investigate the role of the nonstationary dynamics of functional brain activity in behaviour and cognition using a combination of novel data analysis techniques and simultaneous acquisition of functional magnetic resonance imaging (fMRI) and electroencephalography data. The proposed research project is inherently multifaceted and interdisciplinary, straddling the interface between three intersecting fields: data analysis, the physics of neuroimaging methods, and neuroscience. In recent years, there is an increasing interest in the application of resting state fMRI to investigate dynamics of functional networks in health and disease. While most analysis approaches assume temporal stationarity of functional networks, it is becoming more evident that spontaneous activity in functional networks also comprise more dynamic and transient brain states, which could last few minutes or be as short as few seconds as in the time scale of BOLD events. Current efforts aim at characterizing such transient brain states, e.g. in terms of spatial similarity or switching rates across groups of individuals, in order to improve our understanding of the dynamic processes that constitute them. However, grouping across individuals inherently excludes the identification of brain states relating to an individual’s specific cognitive processes as well as those that reflect more intrinsic neuronal fluctuations. As result of the project, novel signal processing and data analysis methods have been developed for the study of the dynamics of spontaneous brain activity in individual subjects. The Paradigm Free Mapping algorithm has been implemented in AFNI, which a widely used, open-source software for the analysis of functional MRI data. The distribution of the algorithms developed as part of the project in AFNI will thus substantially enhance the impact of the project in the scientific community. From a practical point of view, the analysis of resting state fMRI data with 3dPFM enabled us to demonstrate that few transient BOLD events convey sufficient information to reveal synchronous spontaneous activity in functional brain networks (see Figure 1). Consistent mapping of the major functional networks (default mode network, dorsal attention network, visual network, sensorimotor network and executive network) was achievable in individual subjects. We found that spontaneous brain activity extends largely across gray matter areas of the cortex, resembling previous observations that the large regions of the human cortex significantly respond to cognitive tasks. Importantly, the new methods enable to analyze single spontaneous events that occur at rest, thereby their amplitude and timing can be analyzed to quantitatively examine time-varying changes in spontaneous brain connectivity within and between functional networks at short time scales. 3dPFM is also data-adaptive and is rooted on statistical principles so that the number and the amplitude of the detected events can vary according to the temporal characteristics of the signal. These features prevents 3dPFM from setting ad-hoc thresholds in the amplitude of the signal or the number of significant events that must be identified, either in specific ROIs or across the entire brain, in order to investigate the non-stationary dynamics of functional brain connectivity. Additionally, single-subject brain parcellation of functional networks can be achieved if 3dPFM is combined with clustering algorithms that group the spontaneous events detected with 3dPFM. The proposed methods would enable a finer evaluation of between-subject differences in functional connectivity pattern measured with BOLD fMRI and allow the characterization of brain states reflecting cognitive processes and more intrinsic neuronal fluctuations, being potentially useful for the identification of brain disorders. These results have been presented in 2 international conferences:
1-Caballero-Gaudes C, Saad ZS, Raemaekers M, Ramsey NF, Petridou N. Individual subject mapping of functional networks from sparse spontaneous BOLD events. Annual Meeting of the International Society of Magnetic Resonance for Medicine, ISMRM 2015, Toronto, Canada.
2-Caballero-Gaudes C, Saad ZS, Raemaekers M, Ramsey NF, Petridou N. Few spontaneous BOLD events are sufficient for single subject mapping of functional networks at 7T. Annual Meeting of the Organization for Human Brain Mapping, OHBM 2015, Honolulu, Hawaii, USA.
Attached document: Figure 1
The second aim of the project was to investigate the electrophysiological correlates of the transient spatio-temporal clusters of BOLD activations revealed by the analysis of the resting state fMRI signal with 3dPFM. This research entails the development of protocols for, a novel technique at the host institution. During the course of the project, the setup that enables the simultaneous acquisition of EEG and fMRI data has been implemented for the first time at the host institution, with the collaboration of researchers from the Sir Peter Mansfield Imaging Center, University of Nottingham, UK. To achieve this goal, we adapted the algorithms required for an accurate correction of pulse-related artefacts in the EEG signal based on the detection of cardiac pulses on the vectorcardiogram (VCG) signal of the MAGNETOM Trio MR scanner. In parallel, in-house scripts for physiological noise correction of fMRI data have been also developed, including the RETROICOR algorithm (Glover et al., 2000), correction with the Respiration Volume and Cardiac Rate time series (Birn et al., 2006; 2008; Shmueli et al., 2008), and the RVHR method (Chang et al., 2009). These algorithms can work either with physiological signals recorded via the MR scanner’s physiological monitoring unit or via the BIOPAC data acquisition system. Simultaneous EEG-fMRI data while participants remain at rest or during the execution of a visual paradigm, which serves as control state, have been collected for the first time at the host institution and are currently being analyzed. This work will shed light on the interaction between the neural oscillations supporting mental cognitive processes and intrinsic fluctuations that take place at different frequency bands (delta, alpha, beta and gamma) and the transient spontaneous events that are detected in the BOLD fMRI signal with Paradigm Free Mapping. The impact of these results will provide a more complete picture of the origin of spontaneous brain function and the interactions of large-scale intrinsic connectivity networks.
1-Caballero-Gaudes C, Saad ZS, Raemaekers M, Ramsey NF, Petridou N. Individual subject mapping of functional networks from sparse spontaneous BOLD events. Annual Meeting of the International Society of Magnetic Resonance for Medicine, ISMRM 2015, Toronto, Canada.
2-Caballero-Gaudes C, Saad ZS, Raemaekers M, Ramsey NF, Petridou N. Few spontaneous BOLD events are sufficient for single subject mapping of functional networks at 7T. Annual Meeting of the Organization for Human Brain Mapping, OHBM 2015, Honolulu, Hawaii, USA.
Attached document: Figure 1
The second aim of the project was to investigate the electrophysiological correlates of the transient spatio-temporal clusters of BOLD activations revealed by the analysis of the resting state fMRI signal with 3dPFM. This research entails the development of protocols for, a novel technique at the host institution. During the course of the project, the setup that enables the simultaneous acquisition of EEG and fMRI data has been implemented for the first time at the host institution, with the collaboration of researchers from the Sir Peter Mansfield Imaging Center, University of Nottingham, UK. To achieve this goal, we adapted the algorithms required for an accurate correction of pulse-related artefacts in the EEG signal based on the detection of cardiac pulses on the vectorcardiogram (VCG) signal of the MAGNETOM Trio MR scanner. In parallel, in-house scripts for physiological noise correction of fMRI data have been also developed, including the RETROICOR algorithm (Glover et al., 2000), correction with the Respiration Volume and Cardiac Rate time series (Birn et al., 2006; 2008; Shmueli et al., 2008), and the RVHR method (Chang et al., 2009). These algorithms can work either with physiological signals recorded via the MR scanner’s physiological monitoring unit or via the BIOPAC data acquisition system. Simultaneous EEG-fMRI data while participants remain at rest or during the execution of a visual paradigm, which serves as control state, have been collected for the first time at the host institution and are currently being analyzed. This work will shed light on the interaction between the neural oscillations supporting mental cognitive processes and intrinsic fluctuations that take place at different frequency bands (delta, alpha, beta and gamma) and the transient spontaneous events that are detected in the BOLD fMRI signal with Paradigm Free Mapping. The impact of these results will provide a more complete picture of the origin of spontaneous brain function and the interactions of large-scale intrinsic connectivity networks.