One of the research objects in the lab is on the development of deep-learning algorithms for cardiopulmonary MRI processing, especially the development of unsupervised learning algorithms. We are interested in 2D and 3D cardiopulmonary MRI reconstruction, denoising, super-resolution and segmentation. In this objective, we have developed several algorithms such as G-SToRM, MoCo-SToRM, V-SToRM for real-time cardiopulmonary processing. These work on real-time cardiopulmonary processing wothe best paper award in 2020, and the best paper award finalist in 2021 at IEEE International Symposium on Biomedical Imaging (IEEE ISBI). We've also extended the above algorithms for better performances based on recent advances in deep learning such as the inclusion of Spatial Transformer Networks (STN), Implicit Neural Representations (INR), Graph Convolutional Network (GCN),Kernel method into the algorithms.