Human imaging at meso-scale

What is the cortex of the brain?

The outer layer of the brain is the cerebral cortex. It is composed of about 100 billion nerve cells and is called grey matter. Due to its typical folding, it takes up about half of the brain volume. Already at the beginning of the 20th century, the cortex was divided into six layers according to its cell architecture, and due to differences in this architecture into different functional regions. Various studies have been able to demonstrate a correlation between, for example, neurological diseases such as Alzheimer’s and a decrease in the thickness of the cortex in certain functional regions, which can be directly linked to symptoms of the disease, such as memory problems. This results in a direct correlation between the spatial expression of a functional region and the quality of its function.

Tasks in the Collaborative Research Center

The project Z02 plays an important role as a central project in the Collaborative Research Center, as we support the other projects in challenges of image acquisition and analysis using 7 Tesla magnetic resonance imaging (7 T-MRI), and by pushing forward the limits of novel technologies ourselves. Current research on neural resources in humans particularly benefits from novel technology and/or methodology required to describe neural changes at the mesoscale (i.e. less than 1 mm). This benefits the transfer of knowledge from animal research findings (described at the micro level) to brain models and interventions in humans. To this end, we are establishing state-of-the-art MR sequences that provide reproducible and optimized data quality, and computational tools and analysis pipelines for multimodal and multiscale data modelling within and between individuals. In particular, these new methods extend the previous view of the human cortex as a folded surface to consider the depth within the cortex, allowing its different layers to be taken into account in data analysis. As a result, research questions about neural resources can literally be answered in previously undiscovered dimensions.

Artifacts by high field strengths

However, one challenge to successful imaging at the mesoscale of the living human brain is the high sensitivity of these methods to image degradation effects. Indeed, the number and magnitude of image artefacts increase dramatically at field strengths of 7 T compared to 1.5 T and 3 T MRI data (Ladd et al. 2018). Many of these artefacts are caused by subject motion and limit the actual effective image resolution and image quality that can be achieved. In functional MRI (fMRI), brain areas are specifically activated by so-called stimuli. These stimuli can for example be sounds or images. If the participant’s movement correlates with the presentation of the stimulus, the movement can falsify the measurement of brain activations. In addition, the test participant him-/herself is a source of interference that leads to distortions of the strong magnetic field. Although these distortions in themselves contain information (for example used in quantitative susceptibility mapping, QSM), they cause geometric distortions in the image that can affect the analysis or spatial overlay between different data sets. All these effects make signal processing and artefact correction a significant and indispensable area of research in ultra-high field imaging of the brain, especially when submillimeter resolution is the goal. It is therefore a central aspect of Z02 to develop new methods for artefact and motion correction and make them available to researchers in the Collaborative Research Center.

Challenges posed in data analysis

Another challenge for successful mesoscale imaging of the living human brain is the availability of computational tools and algorithms to model fine-grained brain structures and brain functions, such as cortical layers, in different MRI contrasts and their interaction (Kemper et al. 2018, Edwards et al. 2018). This is non-trivial as cortical layers are tiny and not directly visible in in vivo MRI data due to the lack of contrast. Furthermore, different MRI contrasts are often acquired at different image resolutions and viewing angles and analyzed in different software packages sometimes even requiring different computer systems. Interactions and structure-function relationships on the mesoscale can therefore not be readily determined. Z02 focuses on the optimization and further development of automated analysis pipelines that will enable researchers in the Collaborative Research Center to produce reproducible, observer-independent and generalizable research results. We have used different pipelines and toolkits in previous projects to create mesoscale segmentations on ultra-high resolution MR data and combine them with functional data. In addition, we have used novel structural MRI sequences, such as submillimeter quantitative T1 maps, to investigate the relationship between layer-specific cortex structure and the large-scale organization of functional networks (Kuehn et al. 2017). We optimize, complement and validate these methods as part of the Z02 project and make them accessible to researchers in the Collaborative Research Center.

Matching data between individuals

Another major challenge for successful interpretation of mesoscale data is the need to overlay functional and structural images across different individuals. Group statistics are important to investigate the generalizability of effects across individuals and to test for statistically significant group effects, such as comparing older adults with SuperAgers (people whose brain does not seem to age) – a central goal of the Collaborative Research Center. To date, there are very few computational tools or methods optimized for inter-individual matching of mesoscale brain imaging data. In cases where functional areas do not match anatomical landmarks, discrepancies occur. Any small deviation in the location of activation maps or anatomical structures (e.g. Kuehn et al. 2017) can have a massive impact on the outcome of layer-specific analyses, increasing the need for an exact matching procedure. Furthermore, especially in the clinical context, the amount of imaging data that can be obtained from an individual is limited due to procedural challenges, costs and limited compliance. It is therefore the aim of Z02 to enable adequate inter-individual matching through valid data aggregation across individuals, and to make these methods available to other researchers in the Collaborative Research Center.