Extraction of Functional Subnetworks in Autism Using Multimodal MRI

Yale University, New Haven CT

Investigators

Linked publications & trials

Abstract

The Autism spectrum disorders (ASD) are devastating neurodevelopmental disorders with a rising prevalence in the United States that is currently estimated at 1 in 110. The cost of ASD to affected people, families, and society is enormous. Research aimed at uncovering the pathogenesis of this condition, and potentially leading to rational approaches to prevention or treatment, is of the greatest importance. We propose here to use multimodal neuroimaging to identify biomarkers of risk for autism. The discovery of reliable biomarkers will aid in the identification of individuals with ASD as well as those who will subsequently develop or are already developing subtle signs of ASD. In addition, biomarkers could serve to identify early biological risk factors for ASD, ultimately allowing us to achieve the goal of preventing the development of the disorder in people at risk or reducing the degree of severity in those affected. Structural MR image-derived biomarkers related to increased brain volume have revealed differences between ASD and typically-developing comparison participants. Functional (fMRI) differences have also been found (for instance using tasks related to face identity and facial expression as shown by our own group). Recent evidence suggests altered connectivity from both structural (diffusion-based) and functional measures in ASD. However, all of these alterations are quite subtle and the findings have been inconsistent. It is our hypothesis in our proposed work that the use of anatomic and diffusion information to guide and constrain the extraction of fMRI-based functional ASD-related subnetworks whose definition is based on both activation and connectivity maps will provide more sensitive and robust image-derived biomarkers for ASD. Thus, we focus our efforts on the development of a unique mathematical approach that will estimate three functionally-connected subnetworks in the brain related to ASD using a multi-view integration strategy to jointly consider fMRI time course strength/coherence and DTI-based structural paths. This approach will: i.) estimate voxels where activation is most likely in response to biological motion and compute model-enhanced activation regression parameters, ii.) estimate structurally-informed, functionally-connected subnetworks made up of regions of voxels that activate in response to biological motion and iii.) derive important ASD-related parameters of activation signal strength and connectivity that can be used as quantitative biomarkers. We will first apply the strategy to typically developing children and confirm the utility of our measures by illustrating that our approach can produce reliable biomarker information even in the presence of a small number of fMRI runs. Then, to show the effectiveness of the biomarkers that can be derived from our new approach, we will examine signal change and connectivity parameters derived from three ASD-related functional subnetworks that respond to our biological motion task. We will evaluate how well these measures can stratify three subject groups: children with ASD, unaffected siblings of children with ASD and typically developing children. We will then compare the results to three alternative strategies.