Retrospective Motion Correction of Magnitude-Input MR Images
There has been a considerable progress recently in understanding and developing solutions to the problem of image quality deterioration due to patients’ motion in MR scanners. Retrospective methods can be applied to previously acquired motion corrupted data, however, such methods require complex-valued raw volumes as input. It is common practice, though, to preserve only spatial magnitudes of the medical scans, which makes the existing post-processing-based approaches inapplicable. In this work, we make first humble steps towards solving the problem of motion-related artifacts in magnitude-only scans. We propose a learning-based approach, which involves using large-scale convolutional neural networks to learn the transformation from motion-corrupted magnitude observations to the sharp images.
- 1.Zaitsev, M., Maclaren, J., Herbst, M.: Motion artifacts in MRI: a complex problem with many partial solutions. Journal of Magnetic Resonance Imaging (2015)Google Scholar
- 8.Loktyushin, A., Nickisch, H., Pohmann, R., Schölkopf, B.: Blind retrospective motion correction of MR images. Magnetic Resonance in Medicine (2013). doi:10.1002/mrm.24615. (Epub ahead of print)
- 10.Burger, H.C., Schuler, C.J., Harmeling, S.: Image denoising with multi-layer perceptrons, part 1: comparison with existing algorithms and with bounds. CoRR abs/1211.1544 (2012)
- 11.Schuler, C.J., Burger, H.C., Harmeling, S., Schölkopf, B.: A machine learning approach for non-blind image deconvolution. In: Proceedings of the 2013 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2013, pp. 1067–1074. IEEE Computer Society, Washington, DC (2013)Google Scholar
- 12.Schuler, C.J., Hirsch, M., Harmeling, S., Schölkopf, B.: Learning to deblur. CoRR abs/1406.7444 (2014)
- 14.Zeiler, M.D.: ADADELTA: an adaptive learning rate method. CoRR abs/1212.5701 (2012)