Assessment of regional non-linear tissue deformation and air volume change of human lungs via image registration

We evaluate the non-linear characteristics of the human lung via image registration-derived local variables based on volumetric multi-detector-row computed tomographic (MDCT) lung image data of six normal human subjects acquired at three inflation levels: 20% of vital capacity (VC), 60% VC and 80% VC. Local variables include Jacobian (ratio of volume change) and maximum shear strain for assessment of lung deformation, and air volume change for assessment of air distribution. First, the variables linearly interpolated between 20% and 80% VC images to reflect deformation from 20% to 60% VC are compared with those of direct registration of 20% and 60% VC images. The result shows that the linearly-interpolated variables agree only qualitatively with those of registration (P<0.05). Then, a quadratic (or linear) interpolation is introduced to link local variables to global air volumes of three images (or 20% and 80% VC images). A sinusoidal breathing waveform is assumed for assessing the time rate of change of these variables. The results show significant differences between two-image and three-image results (P<0.05). The three-image results for the whole lung indicate that the peak of the maximum shear rate occurs at about 37% of the maximum volume difference between 20% and 80% VC, while the peaks for the Jacobian and flow rate occur at 50%. This is in agreement with accepted physiology whereby lung tissues deform more at lower lung volumes due to lower elasticity and greater compliance. Furthermore, the three-image results show that the upper and middle lobes, even in the recumbent, supine posture, reach full expansion earlier than the lower lobes.