A Joint Network Optimization Framework to Predict Clinical Severity from Resting State Functional MRI Data

TL;DR

A novel optimization framework to predict clinical severity from resting state fMRI (rs-fMRI) data that outperforms standard semi-supervised frameworks and robustly identifies clinically relevant networks characteristic of ASD.

Abstract

We propose a novel optimization framework to predict clinical severity from resting state fMRI (rs-fMRI) data. Our model consists of two coupled terms. The first term decomposes the correlation matrices into a sparse set of representative subnetworks that define a network manifold. These subnetworks are modeled as rank-one outer-products which correspond to the elemental patterns of co-activation across the brain; the subnetworks are combined via patient-specific non-negative coefficients. The second term is a linear regression model that uses the patient-specific coefficients to predict a measure of clinical severity. We validate our framework on two separate datasets in a ten fold cross validation setting. The first is a cohort of fifty-eight patients diagnosed with Autism Spectrum Disorder (ASD). The second dataset consists of sixty three patients from a publicly available ASD database. Our method outperforms standard semi-supervised frameworks, which employ conventional graph theoretic and statistical representation learning techniques to relate the rs-fMRI correlations to behavior. In contrast, our joint network optimization framework exploits the structure of the rs-fMRI correlation matrices to simultaneously capture group level effects and patient heterogeneity. Finally, we demonstrate that our proposed framework robustly identifies clinically relevant networks characteristic of ASD.

Figures (30)

• Figure 1: We group voxels in the brain into ROIs defined by a standard atlas and compute the average time courses for each ROI. The correlation matrix captures the synchrony in the average time courses
• Figure 2: A two level joint model for connectivity and prediction. Purple Box: Depicts the functional data representation or 'generative' term. The correlation matrix is decomposed into a group basis term and a patient specific coefficient term. The columns of the basis matrix correspond to individual subnetworks when projected onto the brain. We stack these coefficients into a matrix. Green Box: Prediction of symptom severity via linear regression
• Figure 3: Our Optimization Strategy, we iterate through four main steps until global convergence
• Figure 4: A typical two stage baseline. We input the correlation matrices to Stage $1$, which performs Feature Extraction on the raw correlations. This step could be a technique from machine learning, graph theory or a statistical measure. Stage $2$ fits an associative regression model to the output representation of Stage $1$
• Figure 5: The graphical model for the joint objective. For our synthetic experiments, we fix the model parameters $\mathbf{\sigma}_{\mathbf{C}}= 2, \mathbf{\sigma}_{\mathbf{w}}= 0.2$