About team BDD
● Mission:
To develop AI models for brain disorder diagnosis that detect disease-specific neurological patterns in resting-state fMRI data
● Scope:
The entire diagnosis pipeline from data augmentation and graph structure learning to cross‑site validation and deployment
● Goal:
To support more precise and robust brain disorder diagnosis with deep learning models
Available internship topics
● Graph-based functional connectivity learning for brain disorder diagnosis
● Domain shift mitigation for improved generalization in multi-site fMRI analysis
Our research topics
Synthetic FC Generation
● We utilize generative models to create topology-aware synthetic Functional Connectivity (FC) matrices for data augmentation and diagnostic enhancement to overcome the scarcity of high-quality neuroimaging data.
Related publications
[JBHI’24] Graph-based Conditional Generative Adversarial Networks for Major Depressive Disorder Diagnosis with Synthetic functional Brain Network Generation
Multi-Atlas Integration
● We tackle the single-scale limitations of functional connectivity analysis by fusing connectivity networks from multiple brain parcellations into a unified representation, enhancing diagnostic accuracy and robustness for neurological disorders.
Related publications
[JBHI'24] Spectral Graph Neural Network-based Multi-atlas Brain Network Fusion for Major Depressive Disorder Diagnosis
Graph Structure Learning
● We leverage graph structure learning to construct subject-specific brain networks that overcome the limitations of fixed correlation measures, enabling accurate and interpretable diagnoses of brain disorders and revealing disorder-related network disruptions for potential biomarker discovery.
Domain Generalization
● We aim to mitigate variability in multi-site fMRI data by reducing site-specific noise while preserving disorder-related patterns, enabling a single model to achieve reliable, high-accuracy brain disorder diagnoses across unseen sites.