Wavelet-Based Iterative Network for Dual-Domain Sparse-View CT Reconstruction Using MRI Priors

Sparse-view CT imaging reduces radiation dose and scan time but often leads to poor image quality. Traditional methods primarily use CT characteristics as priors, overlooking priors from other modalities such as Magnetic Resonance Imaging (MRI). Due to the rich soft tissue information from MRI, it may be beneficial to extract relevant features for CT reconstruction. In this study, we propose a method using wavelet transform to extract features from MRI images as priors for sparse-view CT reconstruction through a dual-domain iterative network. Our approach decomposes the CT image into high-frequency and low-frequency components, and recovers the high-frequency details using MRI-derived priors. This multi-modality data-driven model integrates information from both modalities and across both the image and sinogram domains. Experiments show that our method effectively reduces artifacts and improves image quality, demonstrating the potential of leveraging multi-modality information in enhancing sparse-view CT imaging.