Ingo Hermann1,2, Eloy Martínez-Heras3, Benedikt Rieger1, Ralf Schmidt1, Alena-Kathrin Golla1,4, Jia-Sheng Hong5, Wei-Kai Lee5, Wu Yu-Te5,6, Martijn Nagtegaal2, Elisabeth Solana3, Sara Llufriu3, Achim Gass7, Lothar R Schad1, Sebastian Weingärtner2, Frank G Zöllner1,4. 1. Computer Assisted Clinical Medicine, Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany. 2. Department of Imaging Physics, Delft University of Technology, Delft, the Netherlands. 3. Center of Neuroimmunology, Laboratory of Advanced Imaging in Neuroimmunological Diseases, Hospital Clinic Barcelona, Institut d'Investigacions Biomédiques August Pi i Sunyer (IDIBAPS) and Universitat de Barcelona, Barcelona, Spain. 4. Mannheim Institute for Intelligent Systems in Medicine, Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany. 5. Department of Biomedical Imaging and Radiological Sciences, National Yang-Ming University, Taipei, Taiwan. 6. Institute of Biophotonics and Brain Research Center, National Yang-Ming University, Taipei, Taiwan. 7. Department of Neurology, Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany.
Abstract
PURPOSE: To develop an accelerated postprocessing pipeline for reproducible and efficient assessment of white matter lesions using quantitative magnetic resonance fingerprinting (MRF) and deep learning. METHODS: MRF using echo-planar imaging (EPI) scans with varying repetition and echo times were acquired for whole brain quantification of T 1 and T 2 ∗ in 50 subjects with multiple sclerosis (MS) and 10 healthy volunteers along 2 centers. MRF T 1 and T 2 ∗ parametric maps were distortion corrected and denoised. A CNN was trained to reconstruct the T 1 and T 2 ∗ parametric maps, and the WM and GM probability maps. RESULTS: Deep learning-based postprocessing reduced reconstruction and image processing times from hours to a few seconds while maintaining high accuracy, reliability, and precision. Mean absolute error performed the best for T 1 (deviations 5.6%) and the logarithmic hyperbolic cosinus loss the best for T 2 ∗ (deviations 6.0%). CONCLUSIONS: MRF is a fast and robust tool for quantitative T 1 and T 2 ∗ mapping. Its long reconstruction and several postprocessing steps can be facilitated and accelerated using deep learning.
PURPOSE: To develop an accelerated postprocessing pipeline for reproducible and efficient assessment of white matter lesions using quantitative magnetic resonance fingerprinting (MRF) and deep learning. METHODS: MRF using echo-planar imaging (EPI) scans with varying repetition and echo times were acquired for whole brain quantification of T 1 and T 2 ∗ in 50 subjects with multiple sclerosis (MS) and 10 healthy volunteers along 2 centers. MRF T 1 and T 2 ∗ parametric maps were distortion corrected and denoised. A CNN was trained to reconstruct the T 1 and T 2 ∗ parametric maps, and the WM and GM probability maps. RESULTS: Deep learning-based postprocessing reduced reconstruction and image processing times from hours to a few seconds while maintaining high accuracy, reliability, and precision. Mean absolute error performed the best for T 1 (deviations 5.6%) and the logarithmic hyperbolic cosinus loss the best for T 2 ∗ (deviations 6.0%). CONCLUSIONS: MRF is a fast and robust tool for quantitative T 1 and T 2 ∗ mapping. Its long reconstruction and several postprocessing steps can be facilitated and accelerated using deep learning.
Authors: David Leitão; Rui Pedro A G Teixeira; Anthony Price; Alena Uus; Joseph V Hajnal; Shaihan J Malik Journal: Phys Med Biol Date: 2021-07-26 Impact factor: 3.609
Authors: Ingo Hermann; Alena K Golla; Eloy Martínez-Heras; Ralf Schmidt; Elisabeth Solana; Sara Llufriu; Achim Gass; Lothar R Schad; Frank G Zöllner Journal: BMC Med Imaging Date: 2021-07-08 Impact factor: 1.930