Functional MRI of the Lungs Using Single Breath-Hold and Self-Navigated Ultrashort Echo Time Sequences

To evaluate three-dimensional (3D) ultrashort echo time (UTE) MRI regarding image quality and suitability for functional image analysis using gradient-echo sequences in breath-hold and with self-navigation.

In this prospective exploratory study, 10 patients (mean age, 21 years; age range, 5-58 years; five men) and 10 healthy control participants (mean age, 25 years; age range, 10-39 years; five men) underwent 3D UTE MRI at 3.0 T. Imaging was performed with a prototypical stack-of-spirals 3D UTE sequence during single breath holds (echo time [TE], 0.05 msec) and with a self-navigated "Koosh ball" 3D UTE sequence at free breathing (TE, 0.03 msec). Image quality was rated on a four-point Likert scale. Edge sharpness was calculated. After semiautomated segmentation, fractional ventilation was calculated from MRI signal intensity (FV_SI) and volume change (FV_Vol). The air volume fraction (AVF) was estimated from relative signal intensity (aortic blood signal intensity was used as a reference). Means were compared between techniques and participants. The Wilcoxon signed rank test and Spearman rank correlation were used for statistical analyses.

Image quality ratings were equal for both techniques. However, stack-of-spirals breath-hold UTE was more susceptible to motion and aliasing artifacts. Mean FV_SI was higher during breath hold than at free breathing (mean ± standard deviation in milliliters of gas per milliliters of parenchyma, 0.17 mL/mL ± 0.06 [minimum, 0.07; maximum, 0.34] vs 0.11 mL/mL ± 0.03 [minimum, 0.06; maximum, 0.17], P = .016). Mean FV_SI and FV_Vol were in good agreement (mean difference: at breath hold, -0.008 [95% confidence interval {CI}: 0.007, -0.024]; ρ = 0.97 vs free breathing, -0.004 [95% CI: 0.007, -0.016]; ρ = 0.91). AVF correlated between both techniques (ρ = 0.94).

Breath-hold and self-navigated 3D UTE sequences yield proton density-weighted images of the lungs that are similar in quality, and both techniques are suitable for functional image analysis.Supplemental material is available for this article.© RSNA, 2020.