3D reconstruction research based on segmented sparse MR images for human spine

Medical image segmentation, especially artificial neural network(ANN)-based segmentation, is a popular computer-aided diagnostic method, which is an important foundation for constructing a visual 3D tissue model. However, for some organs such as the spine, the acquisition of Magnetic Resonance (MR) images is often sparse, thus segmented spinal tissue images also have same problem of sparsity, which is not conducive to the implementation of 3D model reconstruction. In order to mitigate the impact of the above problem, we design a filling algorithm for segmented spinal MR images. Further, based on this algorithm, we design a 3D reconstruction method by combining an ANN model. Our proposed method aiming to achieve a useful addition for sparse segmented spinal MR images, which could improve the performance of 3D reconstruction and visualization.