ORIGINAL ARTICLE PURPOSE We aimed to develop a predictive model for lung water content using ultrashort echo time (UTE) magnetic resonance imaging (MRI) and a sponge phantom.
MATERIALS AND METHODSImage quality was preliminarily optimized, and the signalto-noise ratio (SNR) of UTE was compared with that obtained from a three-dimensional fast gradient echo (FGRE) sequence. Four predetermined volumes of water (3.5, 3.0, 2.5, and 2.0 mL) were soaked in cellulose foam sponges 1.8 cm 3 in size and were imaged with UTE-MRI at 1.5 and 3.0 Tesla (T). A multiple echo time experiment (range, 0.1-9.6 ms) was conducted, and the T2 signal decay curve was determined at each volume of water. A three-parameter equation was fitted to the measured signal, allowing for the calculation of proton density and T2*. The calculation error of proton density was determined as a function of echo time. The constants that allowed for the determination of unknown volumes of water from the measured proton density were calculated using linear regression. RESULTS UTE-MRI provided excellent image quality for the four phantoms and showed a higher SNR, compared to that of FGRE. Proton density decreased proportionally with the decreases in both lung water and field strength (from 3.5 to 2.0 mL; proton density range at 1.5 T, 30.5-17.3; at 3.0 T, 84.2-41.5). Minimum echo time less than 0.6 ms at 1.5 T and 1 ms at 3.0 T maintained calculation errors for proton density within the range of 0%-10%. The slopes of the lines for determining the unknown volumes of water with UTE-MRI were 0.12±0.003 at 1.5 T and 0.05±0.002 at 3.0 T (P < 0.0001).
CONCLUSIONIn a sponge phantom imaged at 1.5 and 3.0 T, unknown volumes of water can be predicted with high accuracy using UTE-MRI.