Fast quantitative MRI using controlled saturation magnetization transfer

• Published in: Magnetic resonance in medicine 2019-02, Vol.81 (2), p.907-920
• Main Authors: A.G. Teixeira, Rui Pedro, Malik, Shaihan J., Hajnal, Joseph V.
• Format: Article
• Language: English
• Subjects: Algorithms; Brain - diagnostic imaging; Computer Simulation; DESPOT; Estimates; Full Papers—Imaging Methodology; Healthy Volunteers; Humans; Imaging, Three-Dimensional; JSR; Magnetic fields; Magnetic Resonance Imaging; Magnetic saturation; Magnetization; Magnets; Mathematical models; Models, Theoretical; Monte Carlo Method; Parameter estimation; Phantoms, Imaging; Radio Waves; relaxometry; Reproducibility of Results; Saturation; steady‐state; Tissues; VFA
• ISSN: 0740-3194; 1522-2594
• DOI: 10.1002/mrm.27442

Purpose This study demonstrates magnetization transfer (MT) effects directly affect relaxometry measurements and develops a framework that allows single‐pool models to be valid in 2‐pool MT systems. Methods A theoretical framework is developed in which a 2‐pool MT system effectively behaves as a single‐pool if the RMS RF magnetic field (B1rms{\text{B}}_{1}^{{{\text{rms}}}) is kept fixed across all measurements. A practical method for achieving controlled saturation magnetization transfer (CSMT) using multiband RF pulses is proposed. Numerical, Phantom, and in vivo validations were performed directly comparing steady state (SS) estimation approaches that under correct single‐pool assumptions would be expected to vary in precision but not accuracy. Results Numerical simulations predict single‐pool estimates obtained from MT model generated data are not consistent for different SS estimation methods, and a systematic underestimation of T2 is expected. Neither effect occurs under the proposed CSMT approach. Both phantom and in vivo experiments corroborate the numerical predictions. Experimental data highlights that even when using the same relaxometry method, different estimates are obtained depending on which combination of flip angles (FAs) and TRs are used if the CSMT approach is not used. Using CSMT, stable measurements of both T1 and T2 are obtained. The measured T1 (T1CSMT)) depends on B1rms{\text{B}}_{1}^{{{\text{rms}}}, which is therefore an important parameter to specify. Conclusion This work demonstrates that conventional single pool relaxometry, which is highly efficient for human studies, results in unreliable parameter estimates in biological tissues because of MT effects. The proposed CSMT framework is shown to allow single‐pool assumptions to be valid, enabling reliable and efficient quantitative imaging to be performed.