Our research aims to accelerate Slice Encoding for Metal Artifact Correction (SEMAC) MRI using multi-contrast deep neural networks for patients with degenerative spine diseases. To reduce the scan time of SEMAC, we propose multi-contrast deep neural networks which can produce high SEMAC factor data from low SEMAC factor data. We investigated acceleration in k-space along the SEMAC encoding direction as well as phase encoding direction to reduce the scan time further. To leverage the complementary information of multi-contrast images, we downsampled the data at different k-space positions. The output of multi-contrast SEMAC reconstruction provided great performance for correcting metal artifacts. The developed networks potentially enable clinical use of SEMAC in a reduced scan time with reasonable quality.
5th International Workshop on Machine Learning for Medical Reconstruction, MLMIR 2022, held in conjunction with 25th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2022, pp.44 - 52