Evaluation of Composed Lung Ventilation with 4DCT and Image Registration
K Du1*, J Reinhardt2 , G Christensen3 , K Ding4 , B Zhao3 , J Bayouth1 , (1) University of Wisconsin Madison, Madison, WI, (2) University of Iowa, Iowa City, IA, (3) University of Iowa, Iowa City, IA, (4) Johns Hopkins University, Baltimore, MD
PresentationsMO-A-BRD-5 Monday 7:30AM - 9:30AM Room: Ballroom D
Regional pulmonary function can be derived using four-dimensional computed tomography (4DCT) combined with deformable image registration. However, only peak inhale and exhale phases have been used thus far while the lung ventilation during intermediate phases is not considered. In our previous work, we have investigated the spatiotemporal heterogeneity of lung ventilation and its dependence on respiration effort. In this study, composed ventilation is introduced using all inspiration phases and compared to direct ventilation.Both methods are evaluated against Xe-CT derived ventilation.
Using an in-house tissue volume preserving deformable image registration, unlike the direct ventilation method, which computes from end expiration to end inspiration, Jacobian ventilation maps were computed from one inhale phase to the next and then composed from all inspiration steps. The two methods were compared in both patients prior to RT and mechanically ventilated sheep subjects. In addition, they wereassessed for the correlation with Xe-CT derived ventilation in sheep subjects. Annotated lung landmarks were used to evaluate the accuracy of original and composed deformation field.
After registration, the landmark distance for composed deformation field was always higher than that for direct deformation field (0IN to 100IN average in human: 1.03 vs 1.53, p=0.001, and in sheep: 0.80 vs0.94, p=0.009), and both increased with longer phase interval. Direct and composed ventilation maps were similar in both sheep (gamma pass rate 87.6) and human subjects (gamma pass rate 71.9),and showed consistent pattern from ventral to dorsal when compared to Xe-CT derived ventilation. Correlation coefficient between Xe-CT and composed ventilation was slightly better than the direct method but not significant (average 0.89 vs 0.85, p=0.135).
More strict breathing control in sheep subjects may explain higher similarity between direct and composed ventilation.When compared to Xe-CT ventilation, no significant difference was found for the composed method.
Funding Support, Disclosures, and Conflict of Interest: NIH Grant: R01 CA166703