Yamin Arefeen1, Onur Beker2, Jaejin Cho3, Heng Yu4, Elfar Adalsteinsson1,5,6, Berkin Bilgic3,5,7. 1. Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA. 2. Computer and Communication Sciences, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland. 3. Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, Massachusetts, USA. 4. Department of Automation, Tsinghua University, Beijing, China. 5. Harvard-MIT Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA. 6. Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA. 7. Department of Radiology, Harvard Medical School, Boston, Massachusetts, USA.
Abstract
PURPOSE: To develop a scan-specific model that estimates and corrects k-space errors made when reconstructing accelerated MRI data. METHODS: Scan-specific artifact reduction in k-space (SPARK) trains a convolutional-neural-network to estimate and correct k-space errors made by an input reconstruction technique by back-propagating from the mean-squared-error loss between an auto-calibration signal (ACS) and the input technique's reconstructed ACS. First, SPARK is applied to generalized autocalibrating partially parallel acquisitions (GRAPPA) and demonstrates improved robustness over other scan-specific models, such as robust artificial-neural-networks for k-space interpolation (RAKI) and residual-RAKI. Subsequent experiments demonstrate that SPARK synergizes with residual-RAKI to improve reconstruction performance. SPARK also improves reconstruction quality when applied to advanced acquisition and reconstruction techniques like 2D virtual coil (VC-) GRAPPA, 2D LORAKS, 3D GRAPPA without an integrated ACS region, and 2D/3D wave-encoded imaging. RESULTS: SPARK yields SSIM improvement and 1.5 - 2× root mean squared error (RMSE) reduction when applied to GRAPPA and improves robustness to ACS size for various acceleration rates in comparison to other scan-specific techniques. When applied to advanced reconstruction techniques such as residual-RAKI, 2D VC-GRAPPA and LORAKS, SPARK achieves up to 20% RMSE improvement. SPARK with 3D GRAPPA also improves RMSE performance by ~2×, SSIM performance, and perceived image quality without a fully sampled ACS region. Finally, SPARK synergizes with non-Cartesian, 2D and 3D wave-encoding imaging by reducing RMSE between 20% and 25% and providing qualitative improvements. CONCLUSION: SPARK synergizes with physics-based acquisition and reconstruction techniques to improve accelerated MRI by training scan-specific models to estimate and correct reconstruction errors in k-space.
PURPOSE: To develop a scan-specific model that estimates and corrects k-space errors made when reconstructing accelerated MRI data. METHODS: Scan-specific artifact reduction in k-space (SPARK) trains a convolutional-neural-network to estimate and correct k-space errors made by an input reconstruction technique by back-propagating from the mean-squared-error loss between an auto-calibration signal (ACS) and the input technique's reconstructed ACS. First, SPARK is applied to generalized autocalibrating partially parallel acquisitions (GRAPPA) and demonstrates improved robustness over other scan-specific models, such as robust artificial-neural-networks for k-space interpolation (RAKI) and residual-RAKI. Subsequent experiments demonstrate that SPARK synergizes with residual-RAKI to improve reconstruction performance. SPARK also improves reconstruction quality when applied to advanced acquisition and reconstruction techniques like 2D virtual coil (VC-) GRAPPA, 2D LORAKS, 3D GRAPPA without an integrated ACS region, and 2D/3D wave-encoded imaging. RESULTS: SPARK yields SSIM improvement and 1.5 - 2× root mean squared error (RMSE) reduction when applied to GRAPPA and improves robustness to ACS size for various acceleration rates in comparison to other scan-specific techniques. When applied to advanced reconstruction techniques such as residual-RAKI, 2D VC-GRAPPA and LORAKS, SPARK achieves up to 20% RMSE improvement. SPARK with 3D GRAPPA also improves RMSE performance by ~2×, SSIM performance, and perceived image quality without a fully sampled ACS region. Finally, SPARK synergizes with non-Cartesian, 2D and 3D wave-encoding imaging by reducing RMSE between 20% and 25% and providing qualitative improvements. CONCLUSION: SPARK synergizes with physics-based acquisition and reconstruction techniques to improve accelerated MRI by training scan-specific models to estimate and correct reconstruction errors in k-space.
Authors: Shreyas S Vasanawala; Marcus T Alley; Brian A Hargreaves; Richard A Barth; John M Pauly; Michael Lustig Journal: Radiology Date: 2010-06-07 Impact factor: 11.105
Authors: Mariya Doneva; Peter Börnert; Holger Eggers; Christian Stehning; Julien Sénégas; Alfred Mertins Journal: Magn Reson Med Date: 2010-10 Impact factor: 4.668
Authors: Berkin Bilgic; Borjan A Gagoski; Stephen F Cauley; Audrey P Fan; Jonathan R Polimeni; P Ellen Grant; Lawrence L Wald; Kawin Setsompop Journal: Magn Reson Med Date: 2014-07-01 Impact factor: 4.668
Authors: Chi Zhang; Seyed Amir Hossein Hosseini; Sebastian Weingärtner; Kâmil Uǧurbil; Steen Moeller; Mehmet Akçakaya Journal: PLoS One Date: 2019-10-23 Impact factor: 3.240
Authors: Seyed Amir Hossein Hosseini; Chi Zhang; Sebastian Weingärtner; Steen Moeller; Matthias Stuber; Kamil Ugurbil; Mehmet Akçakaya Journal: PLoS One Date: 2020-02-21 Impact factor: 3.240