自闭症
计算机科学
人工智能
特征(语言学)
生物
表型
模式识别(心理学)
计算生物学
鉴定(生物学)
功能连接
编码
自闭症谱系障碍
一般化
作者
Wenhao Bi,Guangcheng Dongye,Li Zhang
出处
期刊:
[Institute of Electrical and Electronics Engineers]
日期:2026-05-13
卷期号:: 776-781
标识
DOI:10.1109/cscwd68734.2026.11582022
摘要
The diagnosis of Autism Spectrum Disorder (ASD) remains predominantly reliant on subjective behavioral assessments, underscoring the urgent need for objective neuroimaging biomarkers. Conventional neuroimaging methods are often limited to low-order functional connectivity, failing to capture the brain’s complex, multi-scale interactions. To bridge this gap, we propose a novel multi-channel Graph Convolutional Network (GCN) that fuses multi-scale functional connectivity with key phenotypic data. Our model constructs three distinct connectivity graphs—Low-Order FC (LOFC), temporal High-Order FC (tHOFC), and static High-Order FC (aHOFC)—from resting-state fMRI, while integrating age and sex as auxiliary features. Evaluated on the ABIDE dataset, our approach significantly outperforms traditional machine learning baselines in classification accuracy, demonstrating the efficacy of the proposed multi-scale fusion strategy for objective ASD identification.
科研通智能强力驱动
Strongly Powered by AbleSci AI