Perfusion and ventilation filters for Fourier-decomposition MR lung imaging

MR imaging without the use of contrast agents has recently been used for creating perfusion and ventilation functional lung images. The technique incorporates frequency- or wavelet-domain filters to separate the MR signal components. This paper presents a new, subject-adaptive algorithm for perfusion and ventilation filters design. The proposed algorithm uses a lung signal model for separation of the signal components in the frequency domain. Non-stationary lung signals are handled by a short time Fourier transform. This method was applied to sets of 192 and 90 co-registered non-contrast MR lung images measured for five healthy subjects at the rate of 3,33 images per second, using different slice thicknesses. In each case, the resulted perfusion and ventilation images showed a smaller amount of mutual information, when compared to those obtained using the known lowpass/highpass filter approach.