Effect of respiratory motion on free-breathing 3D stack-of-radial liver R2*relaxometry and improved quantification accuracy using self-gating

Purpose To develop an accurate free-breathing 3D liver R2* mapping approach and to evaluate it in vivo. Methods A free-breathing multi-echo stack-of-radial sequence was applied in 5 normal subjects and 6 patients at 3 Tesla. Respiratory motion compensation was implemented using the inherent self-gating signal. A breath-hold Cartesian acquisition was the reference standard. Proton density fat fraction and R2* were measured and compared between radial and Cartesian methods using Bland-Altman plots. The normal subject results were fitted to a linear mixed model (P < .05 considered significant). Results Free-breathing stack-of-radial without self-gating exhibited signal attenuation in echo images and artifactually elevated apparent R2* values. In the Bland-Altman plots of normal subjects, compared to breath-hold Cartesian, free-breathing stack-of-radial acquisitions of 22, 30, 36, and 44 slices, had mean R2* differences of 27.4, 19.4, 10.9, and 14.7 s(-1) with 800 radial views, and they had 18.4, 11.9, 9.7, and 27.7 s(-1) with 404 views, which were reduced to 0.4, 0.9, -0.2, and -0.7 s(-1) and to -1.7, -1.9, -2.1, and 0.5 s(-1) with self-gating, respectively. No substantial proton density fat fraction differences were found. The linear mixed model showed free-breathing radial R2* results without self-gating were significantly biased by 17.2 s(-1) averagely (P = .002), which was eliminated with self-gating (P = .930). Proton density fat fraction results were not different (P > .234). For patients, Bland-Altman plots exhibited mean R2* differences of 14.4 and 0.1 s(-1) for free-breathing stack-of-radial without self-gating and with self-gating, respectively, but no substantial proton density fat fraction differences. Conclusion The proposed self-gating method corrects the respiratory motion bias and enables accurate free-breathing stack-of-radial quantification of liver R2*.