Inelastic Deformable Image Registration (i-DIR): Capturing Sliding Motion through Automatic Detection of Discontinuities

摘要: Deformable image registration (DIR) is an image-analysis method with a broad range of applications in biomedical sciences. Current DIR on computed-tomography (CT) images the lung and other organs under deformation suffer from large errors artifacts due to inability standard methods capture sliding between interfaces, as transformation models cannot adequately handle discontinuities. In this work, we aim at creating novel inelastic deformable (i-DIR) that automatically detects surfaces capable handling discontinuous motion. Our relies introduction regularization term formulation, where characterized shear strain. We validate i-DIR by studying synthetic datasets strong motion, compare its results against two elastic formulations using landmark analysis. Further, demonstrate applicability medical CT registering images. show delivers accurate estimates local strain are similar fields reported literature, do not exhibit spurious oscillatory patterns typically observed methods. conclude locates regions arise dorsal pleural cavity, delivering significantly smaller than traditional