A Novel Hybrid Multimodal Registration for Lung Radiation Therapy Response Evaluation

Abstract

New imaging methods for quantifying physiological characteristics of tumor sub-volumes require robust registration techniques for the comparison of images acquired at multiple time points. A significant obstacle to this is heterogeneous deformability among the structures to be assessed. Applying a rigid transform only can result in gross misalignment of the lung tissue, while a purely deformable algorithm induces unphysical changes in the target volume. We propose a hybrid rigid/deformable registration algorithm that enables a more physically realistic assessment of response in both the target and the normal structures.