1: \begin{abstract}

2: Image corruption by motion artifacts is an ingrained problem in Magnetic Resonance Imaging (MRI).

3: In this work, we propose a neural network-based regularization term to enhance \textit{Autofocusing}, a classic optimization-based method to remove motion artifacts.

4: The method takes the best of both worlds: the optimization-based routine iteratively executes the blind demotion and deep learning-based prior penalizes for unrealistic restorations and speeds up the convergence.

5: We validate the method on three models of motion trajectories, using synthetic and real noisy data.

6: The method proves resilient to noise and anatomic structure variation, outperforming the state-of-the-art demotion methods.

7: 

8: \keywords{Motion Artifacts \and Demotion \and Autofocusing \and MRI }

9: \end{abstract}

10: