Note: Most Internet Explorer 8 users encounter issues playing the presentation videos. Please update your browser or use a different one if available.

Investigating Functional Connectivity in a Large Sample of Children with Autism Spectrum Disorders

Thursday, 2 May 2013: 09:00-13:00
Banquet Hall (Kursaal Centre)
B. Deen1, R. Saxe2 and K. A. Pelphrey3, (1)Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, MA, (2)Brain and Cognitive Sciences, MIT, Cambridge, MA, (3)Child Study Center, Yale University, New Haven, CT
Background: A number of recent studies have investigated functional connectivity in individuals with Autism Spectrum Disorders (ASD), by assessing correlations between fMRI signals from different brain regions.  However, results of these studies have been relatively inconsistent, with reports of both underconnectivity and overconnectivity in ASD, in a range of different functional networks.  It is not yet clear whether these differences resulted from differences in the specific group of ASD participants recruited in each study, differences in preprocessing or analysis methods used, or differences in the regions and networks investigated.  Additionally, many prior studies are subject to potential confounds, including differences in motion across groups, differences in the extent of task-evoked activity, and differences in the effect of global signal removal.

Objectives: To investigate resting-state functional connectivity in a large sample (N=584) of children with and without ASD, using data from the Autism Brain Imaging Data Exchange (ABIDE).  Furthermore, we assessed a number of functional networks, and accounted for the confounds mentioned above.

Methods: A subset of subjects from the ABIDE dataset were chosen based on the following criteria: subjects were under age 18 and had an IQ above 70, and their data had nearly full coverage of the brain and lacked gross image distortions.  Several novel preprocessing techniques were implemented to eliminate potential confounds: 1) pairs of frames with >.5mm of translation or >.5° of rotation between them were removed from the analysis; 2) instead of global mean removal, data were denoised using CompCorr, a PCA-based technique removing signals from white matter and cerebrospinal fluid. After preprocessing, a smaller subset was chosen that matched distributions of several motion measures across groups, excluded subjects with more than 25% of volumes discarded due to motion, and matched groups on age and gender.  Seed-based functional connectivity analyses were performed using seeds from the default mode, control, dorsal attention, motor and visual networks.  Each group-level model accounted for effects of group as well as age, eye status (open or closed), gender, mean translation, mean rotation, and site of data acquisition.

Results: Across networks, functional connectivity maps were highly similar across groups.  Small but significant group differences were observed in the motor and control networks, in both directions.  Substantial effects of motion, age, and site of acquisition were observed for all networks. 

Conclusions: These results indicate that functional connectivity is highly similar in children with and without ASD, calling into question the use of functional connectivity data as evidence for the underconnectivity hypothesis of ASD.  Consistent with prior research, motion was found to have a strong and pervasive effect of functional connectivity estimates, stressing the importance of tightly controlling motion in studies comparing functional connectivity across groups.  Interestingly, large and reliable differences in functional connectivity were observed across sites, indicating that certain acquisition parameters may be better optimized for this measure.

| More