Mapping brain function in living humans has been a topic of intense research for several decades. It is believed that the brain is organized into sets of functionally distinct regions or networks, each performing its own function. Furthermore, these networks seem to exhibit some hierarchy, being comprised of subnetworks, with distinct but related functions that, evidence suggests, are connected both spatially and temporally. It is known that neural pathways exist between them, providing the anatomical link over which these functional relationships are established. However, the structure, spatial extent and dynamics of such brain functional connectivity remain incompletely understood. Evidence from many studies suggests that these networks are always active and continuously interacting, even when the subject is at rest, which implies the ability to identify such connectivity independently from stimulation using specific stimuli. A plethora of different techniques has been employed in investigatiing the brains functional connectivity, including electroencephalography (EEG), magnetoencephalography (MEG) and positron emission tomography (PET), etc. However, most of studies identifying functionally connected resting state networks (RSN) in humans have used resting state functional magnetic resonance imaging (r-fMRI), which measures the spontaneous fluctuations of the blood oxygen level dependent (BOLD) signal during quiet wakefulness.
At the moment large databases of r-fMRI studies have been set up and made available for the large-scale investigation of functional connectivity networks in humans. The vast majority of these include data acquired using scanners working at 1.5 or 3 Tesla (T) field strengths. While such fields provide reliable (and safe) imaging of the human brain, they are limited in terms of sensitivity and spatial resolution. Ultra-high field strength MRI at 7T is rapidly developing, enabling generation of millimeter resolution images with reasonable signal-to-noise ratio (SNR). Only very recently has it become possible to achieve a whole-brain coverage at such fine spatial resolution, with sufficient temporal resolution for fMRI. Generating such data involves multiple challenges related to building and operating such scanners, generating high quality denoised images, avoiding the dominance of physiological noise and overall improvement of both spatial as well as temporal resolution.
Besides these technological challenges, the ability to reliably generate long time series of whole-brain 3D images with millimeter resolution and increasing temporal resolution represents a large-scale network data problem. Many open questions remain related to the analysis of such data, including the essential question of whether the finer resolution may allow one to infer additional connectivity information or structure in the subnetworks. In order to achieve this, sophisticated network analysis techniques are needed to extract relevant functional connectivity measures from the high-dimensional r-fMRI data in such high-resolution imaging. The resulting functional networks can then be further analyzed to determine spatial and temporal correlations, to infer structural motifs as well as dynamic patterns that may shed additional light into our knowledge of the brain.
The three main outcomes of this project will be: (1) new fMRI data of whole-brain activity obtained using ultra-high field r-fMRI, and pre-processing of this data in order to make it suitable for complex network analysis; (2) extraction of functional connectivity patterns across the brain and underlying functional network and improved methods and algorithms for the analysis of functional networks, with emphasis on methods for the identification of recurrent patterns and clusters in scale-free complex networks; and (3) increased understanding of the added value of ultra-high field MRI in the analysis of resting state brain functional connectivity, as compared with data obtained using other methods.
To achieve these ambitious goals three teams with recognized expertise and complementary skills have joined forces. Leading the project is a team from INESC ID with extensive background and experience in the core tasks of data manipulation and modeling and analysis of complex dynamic systems including advanced algorithms for structure detection and characterization in very large networks. Additionally, a researcher from ISR / IST-ID contributes with extensive experience in fMRI acquisition and analysis, including functional connectivity investigation. Finally, a world renowned team from the Harvard Clinical and Translational Science Center, specialized in high and ultra-high field MRI (3T, 7T and above) provides the expertise for high resolution human fMRI data collection and will make the necessary imaging data available and processed for analysis.
The combined expertise of these teams will allow a major improvement in our knowledge of brain connectivity, a challenge that will be one of the most important ones in the next decades.
