The impact of quality control exclusion criteria on functional connectivity in children with neurodevelopmental disorders Open Access

Wang, Liwei (Spring 2020)

Permanent URL:


Background: Autism Spectrum Disorder (ASD) and Attention Deficit Hyperactivity Disorder (ADHD) are two serious neurodevelopmental diseases in the United States with increasing diagnosis rates. Resting-state functional magnetic resonance imaging (rsfMRI) may be a useful tool to characterize the neural underpinnings of neurodevelopmental disorders. Motion can create large artifacts in fMRI, and consequently, data with head motion is often excluded. These excluded images may contribute to selection bias, potentially mischaracterizing these neurodevelopmental disorders compared to typically developing individuals.

Objective: The purpose of this study was to improve estimates of functional connectivity in neurodevelopmental disorders by using detailed phenotypic information to account for non-random missingness in fMRI data in a large sample of elementary school-aged children, where missingness arises from quality control exclusion criteria.

Methods:We analyzed phenotypic information from 758 individuals and assessed the functional connectivity of a neurodevelopmental disease group and a typical developing group based on analyses of 473 children (8-13 years old), which included 119 subjects with ASD, 119 subjects with ADHD, and 235 typical developing subjects. We examined seed-based functional connectivity for a region located in the left posterior cingulate cortex (L-PCC). We applied doubly robust targeted minimum loss-based estimator (DRTMLE) to account for the possible confounding due to the patterns of missingness (driven by motion) being related to the severity of the neurodevelopmental disorders (captured by phenotypic information). We compared the standard analysis ignoring possible confounding with DRTMLE.

Results: First, the non-random missingness in SRS, PANESS, and GAI is triggered by the participants from multiple studies and some of them did not collect SRS,PANESS, and GAI. Second, we found ASD hyperconnectivity between the L-PCC and frontoparietal regions, and this hyperconnectivity was more extensive in DRTMLEthan in the naive approach. We also found regions of ASD-related hypoconnectivity in the temporal lobe in DRTMLE that was not apparent in the naive approach.Differences between ADHD and TD generally resembled those in ASD versus TD but less extensive.Conclusion:Head motion exclusion criteria may lead to biases in the sample of children included in analyses of neurodevelopmental disorders, where more severe pathology may be excluded due to motion. DRTMLE allows the incorporation of phenotypic information to weight data, which may have led to the larger differences between ASD and typically developing observed in the DRTMLE approach versus the naive approach that ignores missingness. This suggests DRTMLE can be used to improve our understanding of neurodevelopmental disorders.

Table of Contents


1 Introduction

2 Data description

2.1 Study population

2.2 Imaging Data 

3 Methods 

3.1 Estimates of functional connectivity accounting for selection bias 

3.2 Simulation description 

3.3 Impact of motion exclusion criteria on selection bias

3.4 Impact of motion exclusion on naive estimates of functional connectivity

4 Results

4.1 Simulations Results

4.2 Impact of motion exclusion criteria on selection bias

4.3 Impact of motion exclusion on naive estimates of functional connectivity

4.4 Estimates of functional connectivity accounting for selection bias

4.4.1 ASD versus TD

4.4.2 ADHD versus TD

5 Discussion


About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
Subfield / Discipline
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files