Emilie Mussard1,2,3, Tom Hilbert1,2,3, Christoph Forman4, Reto Meuli2, Jean-Philippe Thiran2,3, Tobias Kober1,2,3. 1. Advanced Clinical Imaging Technology, Siemens Healthcare AG, Lausanne, Switzerland. 2. Department of Radiology, Lausanne University Hospital and University of Lausanne, Lausanne, Switzerland. 3. LTS5, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland. 4. SHS DI MR DL CARD, Siemens Healthcare GmbH, Erlangen, Germany.
Abstract
PURPOSE: MP2RAGE T1 -weighted imaging has been shown to be beneficial for various applications, mainly because of its good grey-white matter contrast, its B1 -robustness and ability to derive T1 maps. Even using parallel imaging, the method requires long acquisition times, especially at high resolution. This work aims at accelerating MP2RAGE imaging using compressed sensing. METHODS: A pseudo-phyllotactic Cartesian MP2RAGE readout was implemented allowing for flexible reordering and undersampling factors. The sampling pattern was first optimized based on fully sampled data and a compressed sensing reconstruction. Changes in contrast ratios, automated brain segmentation results, and quantitative T1 values were used for benchmarking. In vivo undersampled data from eleven healthy subjects were then acquired using a 4-fold acceleration with the optimized sampling pattern. The resulting images were compared to the standard parallel imaging MP2RAGE protocol by visual inspection and using the above quality metrics. RESULTS: The application of incoherent undersampling and iterative compressed sensing reconstruction on MP2RAGE acquisitions allows for a 57% time reduction (corresponding to 4-fold undersampling with maintained reference lines, TA = 3:35 minutes) compared to the reference protocol using parallel imaging (GRAPPAx3 acceleration, TA = 8:22 minutes) while obtaining images with similar image quality, morphometric (volume differences = [0.07 ± 1.2-3.8 ± 1.9]%) and T1 -mapping outcomes (T1 error = [6 ± 5.1-37 ± 12.3] ms depending on the different structures). CONCLUSION: A whole-brain MP2RAGE acquisition is feasible with compressed sensing in less than 4 minutes without appreciably compromising image quality.
PURPOSE: MP2RAGE T1 -weighted imaging has been shown to be beneficial for various applications, mainly because of its good grey-white matter contrast, its B1 -robustness and ability to derive T1 maps. Even using parallel imaging, the method requires long acquisition times, especially at high resolution. This work aims at accelerating MP2RAGE imaging using compressed sensing. METHODS: A pseudo-phyllotactic Cartesian MP2RAGE readout was implemented allowing for flexible reordering and undersampling factors. The sampling pattern was first optimized based on fully sampled data and a compressed sensing reconstruction. Changes in contrast ratios, automated brain segmentation results, and quantitative T1 values were used for benchmarking. In vivo undersampled data from eleven healthy subjects were then acquired using a 4-fold acceleration with the optimized sampling pattern. The resulting images were compared to the standard parallel imaging MP2RAGE protocol by visual inspection and using the above quality metrics. RESULTS: The application of incoherent undersampling and iterative compressed sensing reconstruction on MP2RAGE acquisitions allows for a 57% time reduction (corresponding to 4-fold undersampling with maintained reference lines, TA = 3:35 minutes) compared to the reference protocol using parallel imaging (GRAPPAx3 acceleration, TA = 8:22 minutes) while obtaining images with similar image quality, morphometric (volume differences = [0.07 ± 1.2-3.8 ± 1.9]%) and T1 -mapping outcomes (T1 error = [6 ± 5.1-37 ± 12.3] ms depending on the different structures). CONCLUSION: A whole-brain MP2RAGE acquisition is feasible with compressed sensing in less than 4 minutes without appreciably compromising image quality.
Authors: Francesco La Rosa; Erin S Beck; Josefina Maranzano; Ramona-Alexandra Todea; Peter van Gelderen; Jacco A de Zwart; Nicholas J Luciano; Jeff H Duyn; Jean-Philippe Thiran; Cristina Granziera; Daniel S Reich; Pascal Sati; Meritxell Bach Cuadra Journal: NMR Biomed Date: 2022-03-31 Impact factor: 4.478