Since its invention more than 20 years ago, functional Magnetic Resonance Imaging (fMRI) has played a central role in cognitive neuroscience. In recent years, ultra-high field MRI scanners (≥ 7 Tesla) have become available, producing data at previously unobtainable quality.
At present however, the full complexity of fMRI data is poorly understood. Our own previous research, as well as that of several other research groups around the world, has shown that only a small percentage of the variance of fMRI time courses is explained by standard analysis techniques. Furthermore, task-related brain activation appears to be much more global and distributed than previously believed. Ultra-high field data show much more detail, making it even more difficult to understand and interpret the full information content.
The aim of our research is to develop new mathematical methods for the analysis of both standard fMRI data (3 Tesla) as well as ultrahigh field fMRI data (≥ 7 Tesla). We aim to harvest as much of the information content of fMRI as possible in order to advance our understanding of human brain function. Specifically, our work has focused on the development of novel techniques for statistical inference and for the analysis of functional brain networks.
1. Predicting task activations from resting state fMRI
Cognitive fMRI research primarily relies on task-averaged responses over many subjects to describe general principles of brain function. Nonetheless, there exists a large variability between subjects that is also reflected in spontaneous brain activity as measured by resting state fMRI (rsfMRI). Leveraging this fact, several recent studies have therefore aimed at predicting task activation from rsfMRI using various machine learning methods within a growing literature on 'connectome fingerprinting'. In reviewing these results, we found lack of an evaluation against robust baselines that reliably supports a novelty of predictions for this task. On closer examination to reported methods, we found most underperform against trivial baseline model performances based on massive group averaging when whole-cortex prediction is considered. Here we present a modification to published methods that remedies this problem to large extent. Our proposed modification is based on a single-vertex approach that replaces commonly used brain parcellations. We further provide a summary of this model evaluation by characterizing empirical properties of where prediction for this task appears possible, explaining why some predictions largely fail for certain targets. Finally, with these empirical observations we investigate whether individual prediction scores explain individual behavioral differences in a task.
2. Statistical inference for fMRI data
One of the principal goals in functional magnetic resonance imaging (fMRI) is the detection of local activation in the human brain. However, lack of statistical power and inflated false positive rates have recently been identified as major problems in this regard. In this study, we propose a non-parametric and threshold-free framework called LISA to address this demand. It uses a non-linear filter for incorporating spatial context without sacrificing spatial precision. Multiple comparison correction is achieved by controlling the false discovery rate in the filtered maps. Compared to widely used other methods, it shows a boost in statistical power and allows to find small activation areas that have previously evaded detection. The spatial sensitivity of LISA makes it especially suitable for the analysis of high-resolution fMRI data acquired at ultrahigh field (≥ 7 Tesla).
3. Predicting intelligence from fMRI data of the human brain in a few minutes of scan time
A number of recent studies have investigated machine learning techniques for predicting individual behaviour from fMRI. Even though encouraging results have been obtained, excessive scan times - especially in resting state fMRI - are a limiting factor.
In this study, we propose a new machine learning algorithm for predicting individual behaviour of healthy human subjects using both resting state (rsfMRI) as well as task-based fMRI (tfMRI). It achieves dimensionality reduction via ensemble learning and partial least squares regression rather than via brain parcellations or ICA decompositions. In addition, it introduces Ricci-Forman curvature as a novel type of edge weight.
As a proof of concept, we focus on predicting fluid, crystallized and general intelligence scores. In a cohort of 390 unrelated test subjects of the Human Connectome Project, we found correlations between the observed and the predicted general intelligence of more than 50 percent in tfMRI, and of around 59 percent when results from two tasks are combined. We compare these results against a benchmark of existing methods that produced correlations below 50 percent in both rsfMRI and tfMRI. We conclude that with novel machine learning techniques applied to tfMRI it is possible to obtain significantly better prediction accuracies at a fraction of the scan time.