High temporal resolution of dynamic MRI using the PSF model with robust reconstruction

A recently developed method known as partially separable functions (PSF) can be used to generate high temporal resolution images for dynamic imaging. And the accurate estimation of parameters is critical for the result in the PSF model. However, the parameters estimation is often interfered by noises during sampling. So the image reconstruction is fragile sometimes. To improve the robustness of reconstruction, this article proposes a new method jointly using robust Principal Component Analysis (RPCA) and modified truncated Singular Value Decomposition (MTSVD) regularization. The preliminary results of in-vivo Cardiac MR (CMR) experiment have shown that the proposed method can improve the robustness to noises when reconstructing dynamic images with high temporal resolution.