Proposed method demonstrates improved MRI image quality using a kernel-aware network and adaptive denoising, indicating potential in clinical diagnostics.
Key Points
The kernel-aware network improves MRI image resolution, resulting in clearer edges and textures.
Experimental results show superior reconstruction quality on IXI and FastMRI datasets compared to mainstream methods.
The dual diffusion model leverages better learning from the blur kernel for effective image reconstruction.
Collaboration between the kernel-aware prediction and denoising networks significantly enhances overall performance.