High-Fidelity Functional Ultrasound Reconstruction via A Visual Auto-Regressive Framework
This work tackles data scarcity issues in fUS imaging, which hampers practical applications in neurovascular mapping, potentially offering incremental improvements in reconstruction quality.
The paper addresses the challenge of data scarcity in functional ultrasound (fUS) imaging for neurovascular mapping, which limits dataset diversity and model fairness, and proposes a visual auto-regressive framework to reconstruct high-fidelity fUS data.
Functional ultrasound (fUS) imaging provides exceptional spatiotemporal resolution for neurovascular mapping, yet its practical application is significantly hampered by critical challenges. Foremost among these are data scarcity, arising from ethical considerations and signal degradation through the cranium, which collectively limit dataset diversity and compromise the fairness of downstream machine learning models.