Intensity-and-Landmark-Driven, Inverse Consistent, B-Spline Registration and Analysis for Lung Imagery

Lung disease is the number three cause of death in America. Measuring local volume deformation from lung image registration may provide a non-invasive approach for detecting and classifying diseases and provide a means for measuring how these diseases respond to intervention. 3D lung CT images contain easily identifiable landmarks such as airway-tree and vascular-tree branch points that can be used for registration and validation. Intensity-based registration methods complement landmark registration methods by providing dense correspondence information since landmarks only provide sparse correspondence information. Intensity-based registration performs best in regions of strong contrast such as between the lung parenchyma and the chest wall, and between the parenchyma and the blood vessels and larger airways. This paper describes a Landmark, Inverse consistent, Tissue volume preserving, B-Spline (LITS) registration algorithm which can be used to measure local lung volume deformation. This method extends the original tissue volume preserving method by adding landmark information and inverse consistency constraint. LITS registration was applied on three subjects to match lung datasets acquired at functional residual capacity and total lung capacity. The registration errors are small compared to the large overall deformations. Sensitivity analysis was performed by changing node spacing of the parameterization and shows that finer B-Spline lattice setting can reveal more details of the feature deformation.