MoRe-3DGSMR: Motion-resolved reconstruction framework for free-breathing pulmonary MRI based on 3D Gaussian representation.

This study presents a self-supervised, motion-resolved reconstruction framework for high-resolution, free-breathing pulmonary magnetic resonance imaging (MRI) using a three-dimensional Gaussian representation (3DGS). The proposed method leverages 3DGS to address the challenges of motion-resolved 3D isotropic pulmonary MRI reconstruction by enabling smooth interpolation across voxels for continuous spatial representation across different respiratory motion states. Pulmonary MRI data are acquired using a golden-angle radial sampling trajectory, with respiratory motion signals extracted from the center of k-space in each radial spoke. Based on the DC self-gated respiratory signal, the k-space data are sorted into multiple respiratory motion states. A 3DGS framework is then applied to reconstruct a reference image volume from the first motion state. Subsequently, a learnable low-rank sparse motion model is trained to estimate deformation vector fields (DVFs) from the canonical state to the other motion states, which are then used to generate images for the remaining motion states by spatially transforming the reference volume. The proposed reconstruction pipeline is evaluated on six subjects and benchmarked against three state-of-the-art MRI motion-resolved reconstruction methods. The experimental findings demonstrate that the proposed framework effectively reconstructs high-resolution, motion-resolved pulmonary MR images. Compared with existing approaches, it achieves superior image quality, as reflected by higher signal-to-noise ratio and contrast-to-noise ratio. The proposed self-supervised 3DGS-based reconstruction method enables pulmonary MRI to be resolved into distinct respiratory states with isotropic spatial resolution, highlighting its promise for clinical pulmonary MRI.