🤖 AI Summary
Functional MRI (fMRI) suffers from inherently limited spatial resolution, and acquiring ground-truth high-resolution data is practically infeasible. To address this, we propose a self-supervised 3D super-resolution method that requires no paired high-resolution labels. Our approach introduces the first self-supervised 3D SR framework incorporating total variation (TV) regularization, enabling deep networks to jointly model fMRI’s spatiotemporal characteristics while simultaneously enforcing structural smoothness and functional map consistency during reconstruction. The proposed TV-guided self-supervised loss effectively suppresses artifacts and preserves anatomical fidelity of activated regions. Experiments demonstrate that, without access to true high-resolution references, our method significantly enhances spatial resolution and achieves functional activation localization accuracy comparable to fully supervised methods—thereby overcoming the long-standing ground-truth data bottleneck in fMRI super-resolution.
📝 Abstract
While functional Magnetic Resonance Imaging (fMRI) offers valuable insights into cognitive processes, its inherent spatial limitations pose challenges for detailed analysis of the fine-grained functional architecture of the brain. More specifically, MRI scanner and sequence specifications impose a trade-off between temporal resolution, spatial resolution, signal-to-noise ratio, and scan time. Deep Learning (DL) Super-Resolution (SR) methods have emerged as a promising solution to enhance fMRI resolution, generating high-resolution (HR) images from low-resolution (LR) images typically acquired with lower scanning times. However, most existing SR approaches depend on supervised DL techniques, which require training ground truth (GT) HR data, which is often difficult to acquire and simultaneously sets a bound for how far SR can go. In this paper, we introduce a novel self-supervised DL SR model that combines a DL network with an analytical approach and Total Variation (TV) regularization. Our method eliminates the need for external GT images, achieving competitive performance compared to supervised DL techniques and preserving the functional maps.