Mass preserving image registration for lung CT

A Hierarchical Scheme for Geodesic Anatomical Labeling of Airway Trees

“…In order to test reproducibility of the expert and automatic labels, the two CT scans of each subject were registered using the image registration method described in [3], and the labeled airway trees were compared in a common coordinate system. Expert 1, Expert 2 and the automatic algorithm reproduced 14.0, 15.1 and 15.2 labels per subject on average.…”

Selecting registration schemes in case of interstitial lung disease follow‐up in CT

“…Up to now, only one study has reported on ILD follow-up monitoring in the context of 2D axial slices of baseline and follow-up scans, subjected to volume registration. 7 Besides ILD monitoring, image registration has received considerable attention in lung CT image analysis applications, such as atlas registration-based segmentation, 8 lung nodule monitoring, [9][10][11][12] emphysema monitoring and estimation of local ventilation and perfusion, 13,14 and motion estimation for radiotherapy planning, 15,16 exploiting various inspiration/expiration protocols. [17][18][19] Multiresolution nonrigid registration, capable for capturing local lung tissue deformations, accounts for a commonly used approach for lung CT registration utilized in all resolution levels, or in combination with rigid transforms applied in low resolution levels for computational time considerations.…”

“…23 Besides the transformation model, the choice of cost function and optimizer accounts for critical selections in a registration scheme. Commonly used cost functions are sumof-squared differences (SSD) and its variants, 14,24 mutual information (MI) and its variants, 7,25,26 normalized mutual information (NMI), 27 and normalized correlation coefficient (NCC), 28,29 while commonly used optimizers considered include variants of gradient descent optimizers 7,14 and quasiNewton. 19 Finally, while most applications adopt the Gaussian pyramid (GP), 7,11,14,30 simple downsampling 19,31 and the Laplacian 18 pyramid have also been adopted.…”

“…32 Such metrics include distance of corresponding landmark points, as well as alignment of anatomical structures such as lung boundary, fissures, and vessel tree center lines. 14,33 The lung overlap error accounts for a coarse registration performance evaluation metric, 32 which however has been commonly used. 11,34,35 Finally, the determinant of Jacobian matrix of the deformation field is used to detect singularities of deformation fields.…”

A Symmetric 4D Registration Algorithm for Respiratory Motion Modeling

“…These registrations {f ij } can be interpolated to obtain a deforming volumetric model. 3D registration algorithms often approximate natural deformation between two shapes through minimizing certain physical deformation energies [1]or geometric smoothness [2,3]. Pairwise 3D registrations have two general limitations.…”