A flexible framework for the design and optimization of water‐excitation RF pulses using B‐spline interpolation

• Published in: Magnetic resonance in medicine Vol. 93; no. 5; pp. 1896 - 1910
• Main Authors: Sieber, Xavier, Romanin, Ludovica, Bastiaansen, Jessica A. M., Roy, Christopher W., Yerly, Jérôme, Wenz, Daniel, Richiardi, Jonas, Stuber, Matthias, Heeswijk, Ruud B.
• Format: Journal Article
• Language: English
• Published: United States Wiley Subscription Services, Inc 01.05.2025
• Subjects: 5D whole‐heart cardiovascular MRI; Adipose Tissue - diagnostic imaging; Adult; Algorithms; Blood vessels; Bone marrow; Bone Marrow - diagnostic imaging; Coronary artery; Coronary vessels; Coronary Vessels - diagnostic imaging; Design optimization; Evolutionary algorithms; Excitation; fat suppression; Female; free‐running; free‐running cardiac imaging; Heart - diagnostic imaging; Humans; Image Processing, Computer-Assisted - methods; Inhomogeneity; Interpolation; Knee - diagnostic imaging; Lipids; Lipids - chemistry; Magnetic Resonance Imaging - methods; Male; numerical optimization; Optimization; Phantoms, Imaging; Radio Waves; RF pulse design; Signal-To-Noise Ratio; Water - chemistry; water‐excitation; WE pulse; 5D whole‐heart cardiovascular MRI; water‐excitation; WE pulse; fat suppression; free‐running cardiac imaging; RF pulse design; free‐running; numerical optimization
• ISSN: 0740-3194; 1522-2594; 1522-2594
• DOI: 10.1002/mrm.30390

Purpose To implement a flexible framework, named HydrOptiFrame, for the design and optimization of time‐efficient water‐excitation (WE) RF pulses using B‐spline interpolation, and to characterize their lipid suppression performance. Methods An evolutionary optimization algorithm was used to design WE RF pulses. The algorithm minimizes a composite loss function that quantifies the fat–water contrast using Bloch equation simulations. In a first study, B‐spline interpolated optimized (BSIO) pulses designed with HydrOptiFrame with durations of 1 and 0.76 ms were generated for 3 T and characterized in healthy volunteers' knees. The femoral bone marrow SNR was compared to that obtained with to 1–1 WE and lipid insensitive binomial off resonant excitation (LIBRE) pulses. In a second study, in the heart at 1.5 T, the water–fat contrast ratio and coronary artery vessel length obtained with a 2.56 ms BSIO pulse was compared to 1–1 WE and LIBRE pulses in free‐running cardiovascular MR. Results The 1 ms BSIO pulse resulted in higher fat suppression and lower contrast ratio (CR) in the bone marrow than the state‐of‐the‐art pulses (4.1 ± 0.2 vs. 4.7 ± 0.4 and 4.4 ± 0.3 for the BSIO, the 1–1 WE and LIBRE respectively, p < 0.05 vs. both) at 3 T. At 1.5 T, the BSIO pulse resulted in a higher blood‐epicardial fat CR (3.8 ± 1.3 vs. 1.6 ± 0.6 and 2.4 ± 1.1 for the BSIO, 1–1 WE and LIBRE, respectively, p < 0.05 vs. both) and longer traceable left coronary artery vessel length (8.7 ± 1.4 cm vs. 7.0 ± 1.0 cm [p = 0.04] and 7.5 ± 1.2 cm [p = 0.09]). Conclusion The HydrOptiFrame framework offers a new opportunity to design WE RF pulses that are robust to B0 inhomogeneity at multiple magnetic field strengths and for variable RF pulse durations.