Abstract:Under the framework of network-based neurodegeneration, brain functional connectome (FC)-based Graph Neural Networks (GNN) have emerged as a valuable tool for the diagnosis and prognosis of neurodegenerative diseases such as Alzheimer's disease (AD). However, these models are tailored for brain FC at a single time point instead of characterizing FC trajectory. Discerning how FC evolves with disease progression, particularly at the predementia stages such as cognitively normal individuals with amyloid deposition or individuals with mild cognitive impairment (MCI), is crucial for delineating disease spreading patterns and developing effective strategies to slow down or even halt disease advancement. In this work, we proposed the first interpretable framework for brain FC trajectory embedding with application to neurodegenerative disease diagnosis and prognosis, namely Brain Tokenized Graph Transformer (Brain TokenGT). It consists of two modules: 1) Graph Invariant and Variant Embedding (GIVE) for generation of node and spatio-temporal edge embeddings, which were tokenized for downstream processing; 2) Brain Informed Graph Transformer Readout (BIGTR) which augments previous tokens with trainable type identifiers and non-trainable node identifiers and feeds them into a standard transformer encoder to readout. We conducted extensive experiments on two public longitudinal fMRI datasets of the AD continuum for three tasks, including differentiating MCI from controls, predicting dementia conversion in MCI, and classification of amyloid positive or negative cognitively normal individuals. Based on brain FC trajectory, the proposed Brain TokenGT approach outperformed all the other benchmark models and at the same time provided excellent interpretability. The code is available at <a class="link-external link-https" href="https://github.com/ZijianD/Brain-TokenGT.git" rel="external noopener nofollow">this https URL</a>

What problem does this paper attempt to address?

The paper aims to address the diagnosis and prognosis of neurodegenerative diseases (such as Alzheimer's disease), particularly in the early stages of the disease (such as mild cognitive impairment or asymptomatic individuals with amyloid deposition) through the analysis of the trajectory of changes in the functional connectome (FC). Currently, most methods based on graph neural networks (GNN) focus only on functional connectome data at a single time point and fail to effectively capture the evolution of the functional connectome over time. To address this shortcoming, the authors propose a new framework—**Brain Tokenized Graph Transformer (Brain TokenGT)**. This framework has the following features: 1. **Graph Invariant and Variant Embedding (GIVE)**: Generates embedding representations for nodes and spatiotemporal edges. 2. **Brain Informed Graph Transformer Readout (BIGTR)**: Enhances the embedding representations and reads them out through a standard transformer encoder. The experimental section demonstrates that Brain TokenGT significantly outperforms other benchmark models in three tasks on two publicly available longitudinal resting-state fMRI datasets (ADNI and OASIS), including the classification of healthy controls and mild cognitive impairment patients, the prediction of the conversion from mild cognitive impairment to dementia, and the classification of amyloid-positive and amyloid-negative normal cognitive individuals. Additionally, this method provides excellent interpretability, which is beneficial for clinical applications.

Beyond the Snapshot: Brain Tokenized Graph Transformer for Longitudinal Brain Functional Connectome Embedding