High-frequency microvascular imaging of rabbit pup brain for contrast-free super-resolution

This study presents our ongoing approach to high-frequency microvascular ultrasound imaging of in vivo rabbit pup brain for contrast-free super-resolution. The method integrates a motion compensation (MC) framework using block matching with non-rigid transformations, along with singular value decomposition filtering, point spread function deconvolution, and morphological operations to enhance signal clarity from microvascular blood flow. Our current results showed that MC naturally reduced misalignment in tissue regions and improves structural similarity across frames. Blood flow signals are also affected by motion, though to a lesser extent. Without compensation, full summation led to directional blurring of blood flow signals, while MC improved the signal-to-noise ratio of peak amplitudes. To further enhance spatial and contrast resolution, we try to incorporate deep learning approaches such as Transformer-based super-resolution models. These models enable fine structural recovery and improved visualization of microvascular features, complementing the MC process. Cine loop comparisons, differential maps, and peak signal-to-noise ratio analysis demonstrate the combined benefits of MC and super-resolution imaging. This integrated approach offers a powerful tool for accurate neurovascular assessment and quantitative analysis in high-resolution ultrasound applications.