Related Experiment Video
Updated: Jun 13, 2025
12:21
Probing the Brain in Autism Using fMRI and Diffusion Tensor Imaging
Published on: September 12, 2011
25.2K
Bilinear Perceptual Fusion Algorithm Based on Brain Functional and Structural Data for ASD Diagnosis and Regions of
Jinxiong Fang1, Da-Fang Zhang2, Kun Xie1
1College of Computer Science and Electronic Engineering, Hunan University, Changsha, 410082, China.
Interdisciplinary Sciences, Computational Life Sciences
|September 10, 2024
Summary
This study introduces a new deep learning method, Bilinear Perceptual Fusion-Graph Convolutional Networks (BPF-GCN), for diagnosing Autism Spectrum Disorder (ASD). BPF-GCN improves ASD classification accuracy by fusing multi-modal brain data, outperforming existing approaches.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Medical Diagnostics
Background:
- Autism Spectrum Disorder (ASD) diagnosis is challenging due to complex mechanisms and varied presentations.
- Current deep learning methods for ASD often rely on single data modalities, limiting information extraction and stability.
Purpose of the Study:
- To develop a novel deep learning framework for improved Autism Spectrum Disorder diagnosis.
- To leverage multi-modal data fusion for more robust ASD classification.
Main Methods:
- Proposed a Bilinear Perceptual Fusion (BPF) algorithm integrating functional and structural brain data.
- Utilized graph convolutional networks (GCNs) to analyze brain network topology and node features.
- Developed the BPF-GCN deep learning framework for ASD classification.
Main Results:
- The BPF-GCN framework achieved a classification accuracy of 82.35% on a public ASD dataset.
- Demonstrated superior performance compared to existing ASD diagnostic methods.
- Successfully identified regions of interest (ROIs) associated with Autism Spectrum Disorder.
Conclusions:
- The BPF-GCN framework offers a significant advancement in the accuracy and stability of Autism Spectrum Disorder diagnosis.
- Multi-modal data fusion using bilinear operations enhances feature representation for ASD detection.
- This approach provides a valuable tool for the timely diagnosis and treatment of ASD.
