Mapping the human connectome at multiple scales with diffusion spectrum MRI

Abstract

The global structural connectivity of the brain, the human connectome, is now accessible at millimeter scale with the use of MRI. In this paper, we describe an approach to map the connectome by constructing normalized whole-brain structural connection matrices derived from diffusion MRI tractography at 5 different scales. Using a template-based approach to match cortical landmarks of different subjects, we propose a robust method that allows (a) the selection of identical cortical regions of interest of desired size and location in different subjects with identification of the associated fiber tracts (b) straightforward construction and interpretation of anatomically organized whole-brain connection matrices and (c) statistical inter-subject comparison of brain connectivity at various scales. The fully automated post-processing steps necessary to build such matrices are detailed in this paper. Extensive validation tests are performed to assess the reproducibility of the method in a group of 5 healthy subjects and its reliability is as well considerably discussed in a group of 20 healthy subjects.

Introduction

The study of neuronal connections in the brain has been a difficult and demanding. Our current knowledge of brain connectivity is largely based on the study of the relationship between symptoms and lesions as well as on post-mortem dissections of large fiber tracts. The former approach was pioneered by Broca (1861) and Wernicke (1906), and the latter by Gall and Spurzheim (1810–1819) and others. More recently, however, great strides have been made in chemical tracing methods in the macaque (Schmahmann and Pandya, 2006) as well as in humans (Clarke et al., 1999, Di Virgilio et al., 1999, Stephan et al., 2008, Zaidel et al., 1995), which have allowed the identification of not only gross fiber tracts but also individual white matter connections. These efforts have resulted in the definitive mapping of a few tens of connections in humans and several hundreds in the macaque. Although such tracing studies are immensely useful and of high-resolution they are also very limited since they are confined to post-mortem material and each study is limited to a few connections only. New high throughput techniques are needed. Important advances have been made with the advent of diffusion MRI tractography, which circumvents the drawbacks mentioned above by allowing not only in vivo (Conturo et al., 1999, Hagmann et al., 2003, Mori et al., 1999, Wedeen, 1996) but also post-mortem imaging (Schmahmann et al., 2007) of a large number of fiber bundles, this, however at the cost of lower resolution. These techniques have spurred many studies related to normal or pathologic neuro-anatomy. More recently it became clear that beyond the aim of characterizing individual fiber bundles, the connectivity profile of the entire brain is of highest importance in neuroscience. Following the pioneering work based on chemical tracing of (Felleman and Van Essen, 1991) and others (Hilgetag and Kaiser, 2004, Hilgetag et al., 2000, Sporns and Zwi, 2004), similar connection matrices have been built from MRI tractography, either by constructing large-scale networks of 1000 nodes (Hagmann et al., 2007) or more anatomically based connection matrices (Gong et al., 2009, Gong et al., 2008, Iturria-Medina et al., 2007, Iturria-Medina et al., 2008, Li et al., 2009, Thottakara et al., 2006). Diffusion-based connectivity has also been used in some studies (Behrens and Johansen-Berg, 2005, Klein et al., 2007, Tomassini et al., 2007) to parcellate gray matter. In order to emphasize the importance of whole brain connectivity, the term connectome has been coined by our group as early as 2005 (Hagmann, 2005, Hagmann et al., 2010, Sporns, 2008, Sporns et al., 2005). It refers to the complete description of the structural connectivity of the brain. More recently our group showed that cortical areas involving the default mode network (Raichle and Snyder, 2007) correspond to highly connected hubs defining the core of structural connectivity (Hagmann et al., 2008). We also showed through computational modeling how low frequency BOLD oscillations can be predicted from structural connectivity, highlighting once again the fundamental relevance of connectomic approaches (Honey et al., 2009).

Given the increasing interest in such approaches as well as emerging questions about the optimal scale and optimal representation of such connection matrices (Bassett et al., 2010, Fornito et al., 2010, Zalesky et al., 2010), we take the opportunity to present in detail our approach to map the connectome at multiple scales and extensively test its reliability and reproducibility.