Brain Effective Connectivity Estimation via Fourier Spatiotemporal Attention

TL;DR

This work tackles brain EC estimation from noisy fMRI by introducing Fourier spatiotemporal attention (FSTA-EC), which combines frequency-domain denoising via a learnable Fourier filter with a unified spatiotemporal attention framework to capture inter-series (spatial) and intra-series (temporal) dependencies. The approach uses FA as an encoder to denoise and extract global frequency features, and STA as a decoder to derive EC $\\mathbf{A}$ and fused spatiotemporal representations, with a loss that promotes sparsity in EC. Empirical results on simulated and real resting-state fMRI show superior performance over state-of-the-art methods in EC recovery and downstream disease classification, while maintaining favorable efficiency. The method offers robust EC estimation under high noise and demonstrates potential for biomarker discovery in neurological and psychiatric contexts.

Abstract

Estimating brain effective connectivity (EC) from functional magnetic resonance imaging (fMRI) data can aid in comprehending the neural mechanisms underlying human behavior and cognition, providing a foundation for disease diagnosis. However, current spatiotemporal attention modules handle temporal and spatial attention separately, extracting temporal and spatial features either sequentially or in parallel. These approaches overlook the inherent spatiotemporal correlations present in real world fMRI data. Additionally, the presence of noise in fMRI data further limits the performance of existing methods. In this paper, we propose a novel brain effective connectivity estimation method based on Fourier spatiotemporal attention (FSTA-EC), which combines Fourier attention and spatiotemporal attention to simultaneously capture inter-series (spatial) dynamics and intra-series (temporal) dependencies from high-noise fMRI data. Specifically, Fourier attention is designed to convert the high-noise fMRI data to frequency domain, and map the denoised fMRI data back to physical domain, and spatiotemporal attention is crafted to simultaneously learn spatiotemporal dynamics. Furthermore, through a series of proofs, we demonstrate that incorporating learnable filter into fast Fourier transform and inverse fast Fourier transform processes is mathematically equivalent to performing cyclic convolution. The experimental results on simulated and real-resting-state fMRI datasets demonstrate that the proposed method exhibits superior performance when compared to state-of-the-art methods.

Figures (5)

• Figure 1: The architecture of FSTA-EC. Specifically, it comprises two elements: (a) Fourier Attention and (b) Spatiotemporal Attention. Fourier attention initially applies fast Fourier transform to convert high-noise fMRI data into frequency domain and then utilizes inverse fast Fourier transform to map the denoised fMRI data back to physical domain. Spatiotemporal attention incorporates temporal attention to learn temporal features from fMRI time series data and employs spatiotemporal fusion attention to simultaneously capture both spatiotemporal dynamic features and spatial brain EC.
• Figure 2: Boxplots of the eight methods for each evaluation metric on the four Sanchez simulated datasets.
• Figure 3: Brain effective connectivity estimated by seven baseline methods and FSTA-EC.
• Figure 4: Visualization of ROIs for real fMRI dataset in the hippocampal head and hippocampal body de2023respective. Abbreviation: CA1 (Cornu Ammonis 1), CA2 (Cornu Ammonis 2), CA3 (Cornu Ammonis 3), DG (Dentate Gyrus), SUB (Subiculum), ERC (Entorhinal Cortex), BA35 (Brodmann Areas 35), BA36 (Brodmann Areas 36), and PHC (Parahippocampal Cortex).
• Figure 5: Hyperparameter analysis on the first simulated fMRI dataset, where the starred results are the best results.

Theorems & Definitions (2)

• Definition 1
• Proposition 1