Attend and Decode: 4D fMRI Task State Decoding Using Attention Models
This work addresses the challenge of decoding brain tasks from high-dimensional fMRI data, which is important for neuroscience and brain-computer interfaces, but it appears incremental as it builds on existing methods like CNNs and attention.
The authors tackled the fMRI task state decoding problem by proposing a novel architecture called Brain Attend and Decode (BAnD), which uses residual CNNs and self-attention for 4D spatio-temporal classification, achieving significant performance gain on a 7-task benchmark from the HCP-YA dataset.
Functional magnetic resonance imaging (fMRI) is a neuroimaging modality that captures the blood oxygen level in a subject's brain while the subject either rests or performs a variety of functional tasks under different conditions. Given fMRI data, the problem of inferring the task, known as task state decoding, is challenging due to the high dimensionality (hundreds of million sampling points per datum) and complex spatio-temporal blood flow patterns inherent in the data. In this work, we propose to tackle the fMRI task state decoding problem by casting it as a 4D spatio-temporal classification problem. We present a novel architecture called Brain Attend and Decode (BAnD), that uses residual convolutional neural networks for spatial feature extraction and self-attention mechanisms for temporal modeling. We achieve significant performance gain compared to previous works on a 7-task benchmark from the large-scale Human Connectome Project-Young Adult (HCP-YA) dataset. We also investigate the transferability of BAnD's extracted features on unseen HCP tasks, either by freezing the spatial feature extraction layers and retraining the temporal model, or finetuning the entire model. The pre-trained features from BAnD are useful on similar tasks while finetuning them yields competitive results on unseen tasks/conditions.