Characterization of regional pulmonary mechanics from serial MRI data

Abstract

We describe a method for quantification of lung motion from the registration of successive images in serial MR acquisitions during normal respiration. MR quantification of pulmonary motion enables in vivo assessment of parenchymal mechanics within the lung in order to assist disease diagnosis or treatment monitoring. Specifically, we obtain estimates of pulmonary motion by summing the normalized cross-correlation over the lung images to identify corresponding locations between the images. The normalized correlation is robust to linear intensity distortions in the image acquisition, which may occur as a consequence of changes in average proton density resulting from changes in lung volume during the respiratory cycle. The estimated motions correspond to deformations of an elastic body and reflect to a first order approximation the true physical behavior of lung parenchyma. The method is validated on a serial MRI study of the lung, for which breath-hold images were acquired of a healthy volunteer at different phases of the respiratory cycle.