Par0056 - elastix

Registration Description

inter and intra-subject; rigid + affine transformation

Image data


For each subject, each baseline scan (time-point 0 wks) was registered to a template scan consisting of two manually drawn labels: one for the right lung and one for the left lung. Those labels were drawn by avoiding the large vessels and the heart (Panel A in Sup-Figure S1). After that, for each subject, the follow-up scans performed post-irradiation (time-points 8, 34, 36, 37, and 40 wks) were registered to the correspondent baseline scan. The template labels were propagated to each scan and the radiodensity values were calculated within those labels. Registration was performed in a coarse-to-fine fashion. Initially, rigid registration was performed to compensate for possible translation and rotation. Afterwards, affine registration was conducted to compensate for possible differences in lung size. A two resolution Gaussian image pyramid was employed in both the rigid and affine registration steps. Mutual information was used as a similarity metric [1].

Registration Settings

For parameter files see the Elastix Model Zoo repository on GitHub.

elastix version: 4.800


A fixed image mask was always used.

Command line call:

elastix -f FixedImage_i.mhd -m MovingImage_j.mhd -fMask FixedImageMask_i -p par0056<X>.txt -out outputdir

with: = one of {rigid, affine}

Published in

[1] D. van Berlo, A. Khmelinskii, A. Gasparini, F. J. Salguero, B. Floot, N. de Wit, M. van de Ven, J. Y. Song, R. P. Coppes, M. Verheij, J. J. Sonke and C. Vens, "Micro Cone Beam Computed Tomography for sensitive assessment of radiation-induced late lung toxicity in preclinical models," accepted, Radiotherapy & Oncology, 2019

© 2020 Viktor van der Valk