Analytical corrections for B1‐inhomogeneity and signal decay in multi‐slice 2D spiral hyperpolarized 129Xe MRI using keyhole reconstruction

• Published in: Magnetic resonance in medicine Vol. 92; no. 3; pp. 967 - 981
• Main Authors: Plummer, J W, Hussain, R, Bdaiwi, A S, Costa, M L, Willmering, M M, J Parra‐Robles, Cleveland, Z I, Walkup, L L
• Format: Journal Article
• Language: English
• Published: Hoboken Wiley Subscription Services, Inc 01.09.2024
• Subjects: Abnormalities; Bias; Cystic fibrosis; Decay; Image acquisition; Image reconstruction; Inhomogeneity; Keyholes; Magnetic resonance imaging; Oversampling; Rescaling; Respiratory function; Root-mean-square errors; Ventilation; Xenon; Xenon 129
• ISSN: 0740-3194; 1522-2594; 1522-2594
• DOI: 10.1002/mrm.30028

PurposeHyperpolarized xenon MRI suffers from heterogeneous coil bias and magnetization decay that obscure pulmonary abnormalities. Non‐physiological signal variability can be mitigated by measuring and mapping the nominal flip angle, and by rescaling the images to correct for signal bias and decay. While flip angle maps can be calculated from sequentially acquired images, scan time and breath‐hold duration are doubled. Here, we exploit the low‐frequency oversampling of 2D‐spiral and keyhole reconstruction to measure flip angle maps from a single acquisition.MethodsFlip angle maps were calculated from two images generated from a single dataset using keyhole reconstructions and a Bloch‐equation–based model suitable for hyperpolarized substances. Artifacts resulting from acquisition and reconstruction schemes (e.g., keyhole reconstruction radius, slice‐selection profile, spiral‐ordering, and oversampling) were assessed using point‐spread functions. Simulated flip angle maps generated using keyhole reconstruction were compared against the paired‐image approach using RMS error (RMSE). Finally, feasibility was demonstrated for in vivo xenon ventilation imaging.ResultsSimulations demonstrated accurate flip angle maps and B1‐inhomogeneity correction can be generated with only 1.25‐fold central‐oversampling and keyhole reconstruction radius = 5% (RMSE = 0.460°). These settings also generated accurate flip angle maps in a healthy control (RSME = 0.337°) and a person with cystic fibrosis (RMSE = 0.404°) in as little as 3.3 s.ConclusionRegional lung ventilation images with reduced impact of B1‐inhomogeneity can be acquired rapidly by combining 2D‐spiral acquisition, Bloch‐equation–based modeling, and keyhole reconstruction. This approach will be especially useful for breath‐hold studies where short scan durations are necessary, such as dynamic imaging and applications in children or people with severely compromised respiratory function.