• The aim is the use of DL methods to make MRI scans faster. By producing accurate images from under-sampled data.
• Developing a Hybrid GAN to reconstruct MR image from 4X and 8X Under Sampled K-Space data from a single-coil and multi-coil MRI. The GAN uses self attention module with dual discriminators to maintain structural integrity during reconstruction and lots of small functionalities to improve reconstruction. More ongoing experiments and ideas.
• Experimenting with a new dehazing architechure for noise removal.
Code is inspired from : https://github.com/KupynOrest/DeblurGAN