A lung 3D model reconstruction method based on compressed sensing and MRI

Lung magnetic resonance imaging (MRI) using hyperpolarized ¹²⁹Xe as contrast agent is an emerging medical imaging technique for respiratory disease diagnosis and therapy evaluation. 3D model can help doctors to have more insights of patients´ lung in disease diagnosis and surgical planning. Conventional 3D reconstruction requires a large number of scanning layers. Different from conventional proton MRI, polarization of hyperpolarized ¹²⁹Xe cannot be restored in the process of scanning. Therefore signal intensity decreases rapidly with the increase of scanning layer numbers. In the meanwhile, image quality suffers from movement artifacts related with multiple breath which also caused by too many scanning layers. In this paper, based on the compressibility of pulmonary 3D spatial structure information, 3D lung model reconstruction is achieved with a few orthogonal scans through compressed sensing (CS) technique. simulation and experimental results show that good quality pulmonary 3D model can be obtained when the number of scanning layers is only 27.3% of full sampling.