Matrix pencil decomposition of time-resolved proton MRI for robust and improved assessment of pulmonary ventilation and perfusion.

Abstract

To present an improved and robust method of pulmonary function assessment from time-resolved proton MRI using a matrix pencil (MP) method in combination with a linear least squares analysis. Simulations of the signal time course in lung parenchyma were performed to compare the accuracy of Fourier decomposition (FD) and MP methods for the estimation of respiratory and cardiac amplitudes. Series of two-dimensional time-resolved lung images were acquired in healthy volunteers at 1.5 T using ultra-fast steady-state free precession. Qualitative lung ventilation- and perfusion-weighted images as well as a quantitative map of fractional ventilation, perfusion, and blood arrival time were calculated using the proposed MP method and compared with the contemporary FD technique. A region-of-interest analysis was performed on the quantitative data. The signal analysis performed using MP decomposition resulted in reduced variability of the estimated respiratory and cardiac amplitudes in comparison with FD for both simulated and in vivo data. MP decomposition provides an automatic, robust, and more accurate estimation of amplitudes of respiratory and cardiac signal modulations in the lung parenchyma than the contemporary FD technique. Magn Reson Med 77:336-342, 2017. © 2016 Wiley Periodicals, Inc.