Multimodal Patterns Associated with Social Interaction Through Gray and White Matter in Autism
Poster No:
448
Submission Type:
Abstract Submission
Authors:
Long Wei1, xin xu1, Suyu Zhong2
Institutions:
1shandong jianzhu university, jinan, shandong, 2Beijing university of posts and telecommunications, Beijing, Beijing
First Author:
Co-Author(s):
Introduction:
Autism spectrum disorder (ASD) patients are characterized by deficits in various aspects of social interaction. Based magnetic resonance imaging (MRI) data, multimodal fusion methods [1,2] are widely applied to explore brain regions related to social impairment in ASD but the role of white matter in brain function has always been underestimated. Using gray and white matter data can help to understand how ASD patients differ from healthy control (HC) in information processing and transmission. So, we have improved the data preparation step to extract purer gray and white matter information and explore multimodal neuroimaging patterns associated with social impairment.
Methods:
699 male participants (ASD/HC: 343/356) from Autism Brain Imaging Data Exchange(ABIDE)Ⅰ&Ⅱ were included, who have total social responsiveness scale (SRS) scores, T1-weighted structural-MRI (sMRI) and resting-state functional MRI (rs-fMRI). sMRI and rs-fMRI data were preprocessed by pipeline tools CAT12 [3] and DPARSF [4] respectively. More specifically, white matter signals were not regressed out and controlled maximum head motion less than 5mm or 5° during the rs-fMRI preprocessing stage. All participants calculated gray matter volume (GMV) map from sMRI and fractional amplitude of low-frequency fluctuations retaining white matter signals (WM-fALFF) map from rs-fMRI. Then, overlaid the white matter mask file onto all processed WM-fALFF maps to save the white matter tissue data only. Finally, we converted all kinds of maps into Z-score maps and obtained two matrices by reshaping each subject's map into a row of vectors and stacked sequentially for each modality. Meanwhile, we concatenated total SRS scores into one-dimensional vectors according to the same sequence. After data preparation, a supervisory data-driven analysis method called "multi-site canonical correlation analysis with reference + joint independent component analysis" (MCCAR +jICA) [5] was used, which can identify joint multimodal component more relevanted to the social interactions by setting total SRS scores as reference data based on two modal matrices (Figure 1).
Results:
Figure 2A showed the spatial maps of identified independent component（ICs）for two modality. For GMV, the identified brain regions mainly located in bilateral insula, bilateral caudate and bilateral hippocampus, accompanied with WM-fALFF mainly located in bilateral corpus callosum, right internal capsule, left inferior longitudinal fasciculus and fornix. Figure 2B indicated that ICs were significantly(p < 0.0005) positive correlations with total SRS scores on all modalities (GMV: r = 0.127, p = 7.33x10-6; WM-fALFF: r = 0.164, p = 1.32x10-5). The higher loadings, the worse social function. We conducted two sample t-test on IC loadings between ASD and HC to explore the group difference (as shown in Figure 2C). The ICs differs significantly across all modalities, and ASD had higher means (GMV: t= 3.52 ,p = 0.0005; WM-fALFF t= 4.60 ,p = 4.92x10-6).
Conclusions:
Our study found a significant gray and white matter function pattern associated with social impairment in ASD, whose brain regions related to information integration, emotional control, language expression, physical movement and external stimuli response. These foundings might provide potential insights to study the causes of social behavior disorders in ASD patients.
Disorders of the Nervous System:
Neurodevelopmental/ Early Life (eg. ADHD, autism) 1
Emotion, Motivation and Social Neuroscience:
Social Interaction 2
Keywords:
Autism
Social Interactions
White Matter
Other - multimodal fusion
1|2Indicates the priority used for review
Provide references using author date format
[2] Qi, Shile et al.(2020), ‘Common and unique multimodal covarying patterns in autism spectrum disorder subtypes’, Molecular autism, vol. 11, no.1, pp. 90.
[3] Gaser C et al.(2022), ‘A Computational Anatomy Toolbox for the Analysis of Structural MRI Data’, bioRxiv.
[4] Yan, C.G et al.(2016), ‘DPABI: Data Processing & Analysis for (Resting-State) Brain Imaging’, Neuroinformatics, vol. 14, no. 3, pp. 339-351.
[5] Li T et al.(2019), ‘Multimodal neuroimaging patterns associated with social responsiveness impairment in autism: A replication study’, 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019), pp. 409-413.