It was recently shown by Margulies et al (2016) that the principal gradients of cortical connectivity can be situated spatially along the default mode network. In addition these principal gradient appear to almost linearly follow a hierarchy in terms of levels of cognitive processing. Autism is a developmental condition that has often been speculated to involve alteration in connectivity and in addition has been shown to effect different cognitive processes at different levels. As such, in the present project we sought to explore the use of cortical connectivity gradients to assess potential differences in cortical hierarchy in individuals with autism.
The relationship between cortical topography in the brain and underlying cognitive processes is complex. Functional magnetic resonance imaging (FMRI) and lesion studies have associated several regions and networks of regions with distinct neural functions. Research on the systematic interaction of brain structure and function on a global scale, however, is scarce. Reason for optimism comes from exciting new evidence revealing the existence of spatial gradients across the brain. These spatial gradients mirror functional processing gradients. More specifically, Margulies et al. (Margulies et al. 2016) identified a principal gradient stretching from anatomical regions associated with primary sensory-motor functions to several regions associated with higher-order processing, such as regions of the Default Mode Network (DMN). This research, conducted in control subjects, suggests that in the healthy brain, the primary gradient can serve as the basis for a principle of cortical organization.
This could be of crucial importance in understanding neurodevelopmental conditions known to result in atypical cortical organisation, such as Autism Spectrum Condition (ASC). ASC is characterised by a multitude of local and global disruptions of functional connectivity (FC) (Vissers et al. 2012). These potential anomalies in connectivity are accompanied by a range of congitive-behavioural anomalies, such as atypical socio-emotional, motion, or language processing. A range of hypotheses have attempted to explain these disruptions in FC and their impact on brain function, such as the hypothesis of reduced long-distance (global) and increased short-distance (local) FC (Belmonte 2004), the ‘underconnectivity hypothesis’ (Just 2004) and more cognitive theories focused on general weakness of central coherence (Happé 2013). While these theories can explain local/regional anomalies in specific ASD subgroups, they frequently conflict and fail to reflect global FC disruptions, and their relation to function, across the ASD spectrum. This necessitates a systematic whole-brain approach to investigate the relationship between FC across the brain and underlying cognitive processes.
Connectivity gradients (Margulies et al. 2016) might provide a way to quantify if alterations in cognitive processing also lead to differences in connectivity hierarchy. Hence, we set out to recreate the gradients from the orginal paper by Margulies et al. (2016) in a publically available and pre-processed autism imaging dataset (http://preprocessed-connectomes-project.org/abide/) and investigate potential differences in the steepness of the gradient, its overlap with the DMN and its overlap with maps corresponding to the associated hierarchical cognitive processing as reported in figure 4 of Margulies et al. (2016).
In order to make the project practically feasible in the course of a brainhack, we utilized all male adult data (age range: 18-55) that was already pre-processed from the ABIDE dataset (http://preprocessed-connectomes-project.org/abide/). We chose to use the most fine-grained parcellation available (Craddock 400) that was pre-processed with the C-PAC pipeline and did not include global signal regression (Craddock et al. 2011). The main processing pipeline for obtaining the gradient components can be found on the Autism Research Centre github repository and speficially gradient ipython notebook. In short, we selected only subjects sthat had no parcels with missing time-series information to avoid dimension mismatches when creating the correlation matrix. This resulted in a total of 178 subjects; 100 individuals with autism (age: 25.01±5.8) and 78 neurotypical controls (25.71±7.42), matched for age (p = 0.53). We then ran diffusion embedding on the thresholded (>10%) matrix and backprojected the diffusion components for each individual subject using the pySTATIS. Resulting gradients files were next written out in nifti format for the primary gradient for every individual. Our primary outcome measure was the slope of the gradient as estimated by the linear fit to the sorted gradient values. As a secondary output we computed a goodness of fit ratio for the gradient values inside and outside of brain masks obtained from NeuroSynth that accompanied the keywords listed in figure 4 of Margulies et al. 2016.