Abstract:Using multimodal neuroimaging data to characterize brain network is currently an advanced technique for Alzheimer's disease(AD) Analysis. Over recent years the neuroimaging community has made tremendous progress in the study of resting-state functional magnetic resonance imaging (rs-fMRI) derived from blood-oxygen-level-dependent (BOLD) signals and Diffusion Tensor Imaging (DTI) derived from white matter fiber tractography. However, Due to the heterogeneity and complexity between BOLD signals and fiber tractography, Most existing multimodal data fusion algorithms can not sufficiently take advantage of the complementary information between rs-fMRI and DTI. To overcome this problem, a novel Hypergraph Generative Adversarial Networks(HGGAN) is proposed in this paper, which utilizes Interactive Hyperedge Neurons module (IHEN) and Optimal Hypergraph Homomorphism algorithm(OHGH) to generate multimodal connectivity of Brain Network from rs-fMRI combination with DTI. To evaluate the performance of this model, We use publicly available data from the ADNI database to demonstrate that the proposed model not only can identify discriminative brain regions of AD but also can effectively improve classification performance.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to effectively utilize multi - modal neuroimaging data (such as resting - state functional magnetic resonance imaging (rs - fMRI) and diffusion tensor imaging (DTI)) to generate multi - modal connectivity of brain networks in Alzheimer's disease (AD) analysis. Specifically, due to the heterogeneity and complexity between BOLD signals and fiber - tract imaging, most existing multi - modal data fusion algorithms cannot fully utilize the complementary information between rs - fMRI and DTI. To overcome this problem, the paper proposes a new hypergraph - based generative adversarial network (HGGAN) to generate multi - modal connectivity of brain networks through the Interactive Hyper - Edge Neuron (IHEN) module and the Optimal Hypergraph Homomorphism (OHGH) algorithm. ### Main Contributions 1. **New Model**: The HGGAN model is proposed, which can more effectively capture the complex relationships between rs - fMRI and DTI. 2. **Improved Module**: The IHEN module is designed, which improves the recent work HNHN, enabling the model to learn the internal relationships and complementary information in hypergraph - structured multi - modal imaging data. 3. **Optimization Algorithm**: The OHGH algorithm is proposed for constructing hypergraph - structured data, which significantly improves the robustness of the generation results. 4. **Experimental Verification**: Experiments were carried out using public data in the ADNI database, proving that this model can not only identify the key brain regions of AD, but also effectively improve the classification performance. ### Method Overview - **Data Pre - processing**: Extract BOLD time series from rs - fMRI data and construct structural connectivity (SC) from DTI data. - **Hypergraph Construction**: Calculate the initial hypergraph using the Dynamic Hypergraph Construction (DHC) algorithm and optimize hypergraph homomorphisms through the OHGH algorithm. - **Generator and Discriminator**: The generator uses the IHEN module to generate a multi - modal connectivity matrix, and the discriminator compares the distribution differences between the functional connectivity (FC) matrix and the multi - modal connectivity matrix through the random walk method. - **Performance Evaluation**: The prediction performance of the model was evaluated through six binary - classification experiments (including AD vs. NC, AD vs. EMCI, etc.), and the results show that using the proposed multi - modal connectivity matrix (MC) significantly improves the classification accuracy. ### Experimental Results - **Classification Performance**: In multiple classification tasks, the classification accuracy using the proposed MC matrix is significantly higher than that using only the structural connectivity (SC) or functional connectivity (FC) matrix. - **Key Brain Region Identification**: By comparing the average correlation coefficients, several key brain regions highly correlated with AD were identified, and these brain regions are mainly concentrated in the memory and reasoning areas. ### Conclusion The method proposed in this paper successfully extracts the complementary information in different modal data, generates individualized multi - modal connectivity matrices, and significantly improves the diagnostic performance of AD. Although this research mainly focuses on AD, the proposed model can be easily extended to the research of other neurodegenerative diseases.

Characterization Multimodal Connectivity of Brain Network by Hypergraph GAN for Alzheimer's Disease Analysis

Regional Brain Fusion: Graph Convolutional Network for Alzheimer's Disease Prediction and Analysis

Interpretable Graph Convolutional Network Of Multi-Modality Brain Imaging For Alzheimer’s Disease Diagnosis

Multi-modal Graph Neural Network for Early Diagnosis of Alzheimer's Disease from sMRI and PET Scans

BG-GAN: Generative AI Enables Representing Brain Structure-Function Connections for Alzheimer's Disease

Hypergraph Structural Information Aggregation Generative Adversarial Networks for Diagnosis and Pathogenetic Factors Identification of Alzheimer’s Disease With Imaging Genetic Data

Multimodal Hyperbolic Graph Learning for Alzheimer's Disease Detection

Multi-Modal Diagnosis of Alzheimer's Disease using Interpretable Graph Convolutional Networks

Alzheimer's Disease Prediction via Brain Structural-Functional Deep Fusing Network

Multimodality Alzheimer's Disease Analysis in Deep Riemannian Manifold

Discovering Brain Network Dysfunction in Alzheimer's Disease Using Brain Hypergraph Neural Network

A Heterogeneous Graph Neural Network Fusing Functional and Structural Connectivity for MCI Diagnosis

Cross-Modality Translation with Generative Adversarial Networks to Unveil Alzheimer's Disease Biomarkers

An Extensible Hierarchical Graph Convolutional Network for Early Alzheimer's Disease Identification

Dynamic Multi-Task Graph Isomorphism Network for Classification of Alzheimer’s Disease

An Extensible Hierarchical Graph Convolutional Network for Early Alzheimer’s Disease Identification

Early Diagnosis and Biomarkers of Alzheimer's Disease Based on Spatio-temporal Graph Convolution Network

MLC-GCN: Multi-Level Generated Connectome Based GCN for AD Analysis

A Self-guided Multimodal Approach to Enhancing Graph Representation Learning for Alzheimer's Diseases

Cross-Modal Transformer GAN: A Brain Structure-Function Deep Fusing Framework for Alzheimer's Disease

Multimodal Neuroimaging Data in Early Detection of Alzheimer's Disease: Exploring the Role of Ensemble Models and GAN Algorithm