Bayesian Optimisation of Large-Scale Biophysical Networks

The relationship between structure and function in the human brain is well established, but not yet well characterised. Large-scale biophysical models allow us to investigate this relationship, by leveraging structural information (e.g. derived from diffusion tractography) in order to couple dynamical models of local neuronal activity into networks of interacting regions distributed across the cortex. In practice however, these models are difficult to parametrise, and their simulation is often delicate and computationally expensive. This undermines the experimental aspect of scientific modelling, and stands in the way of comparing different parametrisations, network architectures, or models in general, with confidence. Here, we advocate the use of Bayesian optimisation for assessing the capabilities of biophysical network models, given a set of desired properties (e.g. band-specific functional connectivity); and in turn the use of this assessment as a principled basis for incremental modelling and model comparison. We adapt an optimisation method designed to cope with costly, high-dimensional, non-convex problems, and demonstrate its use and effectiveness. Using five parameters controlling key aspects of our model, we find that this method is able to converge to regions of high functional similarity with real MEG data, with very few samples given the number of parameters, without getting stuck in local extrema, and while building and exploiting a map of uncertainty defined smoothly across the parameter space. We compare the results obtained using different methods of structural connectivity estimation from diffusion tractography, and find that one method leads to better simulations.

High resolution whole brain diffusion imaging at 7T for the Human Connectome Project

Mapping structural connectivity in healthy adults for the Human Connectome Project (HCP) benefits from high quality, high resolution, multiband (MB)-accelerated whole brain diffusion MRI (dMRI). Acquiring such data at ultrahigh fields (7T and above) can improve intrinsic signal-to-noise ratio (SNR), but suffers from shorter T2 and T2(⁎) relaxation times, increased B1(+) inhomogeneity (resulting in signal loss in cerebellar and temporal lobe regions), and increased power deposition (i.e. specific absorption rate (SAR)), thereby limiting our ability to reduce the repetition time (TR). Here, we present recent developments and optimizations in 7T image acquisitions for the HCP that allow us to efficiently obtain high quality, high resolution whole brain in-vivo dMRI data at 7T. These data show spatial details typically seen only in ex-vivo studies and complement already very high quality 3T HCP data in the same subjects. The advances are the result of intensive pilot studies aimed at mitigating the limitations of dMRI at 7T. The data quality and methods described here are representative of the datasets that will be made freely available to the community in 2015.

Effects of image reconstruction on fiber orientation mapping from multichannel diffusion MRI: reducing the noise floor using SENSE

PURPOSE

To examine the effects of the reconstruction algorithm of magnitude images from multichannel diffusion MRI on fiber orientation estimation.

THEORY AND METHODS

It is well established that the method used to combine signals from different coil elements in multichannel MRI can have an impact on the properties of the reconstructed magnitude image. Using a root-sum-of-squares approach results in a magnitude signal that follows an effective noncentral-χ distribution. As a result, the noise floor, the minimum measurable in the absence of any true signal, is elevated. This is particularly relevant for diffusion-weighted MRI, where the signal attenuation is of interest.

RESULTS

In this study, we illustrate problems that such image reconstruction characteristics may cause in the estimation of fiber orientations, both for model-based and model-free approaches, when modern 32-channel coils are used. We further propose an alternative image reconstruction method that is based on sensitivity encoding (SENSE) and preserves the Rician nature of the single-channel, magnitude MR signal. We show that for the same k-space data, root-sum-of-squares can cause excessive overfitting and reduced precision in orientation estimation compared with the SENSE-based approach.

CONCLUSION

These results highlight the importance of choosing the appropriate image reconstruction method for tractography studies that use multichannel receiver coils for diffusion MRI acquisition.