Par0060 - elastix

Registration Description

Intapatient, rigid+ affine+ b-spline transformation, mutual information

Image data

2D cine-MRI acquired at 1.5T with 50-300 frames ('dynamics') per imaging session.
Imaging parameters: 2D sagittal balanced steady-state free precession sequence (bSSFP). Relaxation time: 2.8 ms, echo time: 1.4 ms, flip angle: 50 degrees, FOV: 320x320 mm, voxel spacing: 1.42x.1.42 mm, slice thickness: 7 mm, readout direction: feet-head, bandwidth: 724-2034 Hz/px. Temporal resolution of each dynamic was 500-570 ms;

Application

The registration was used to align organs in the abdomen of sequential cine-MR dynamics to the reference dynamic, randomly chosen from the cine after entering the steady-state. This registration was used to obtain deformable intrafraction motion from 2D cine-MR images as a potential deep learning ground-truth to learn deformtion vector fields.

Registration Settings

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

elastix version: 4.800

Command line call:

elastix -f  -m  -p param_rigid.txt -p param_affine.txt -p param_bspline.txt -out

where is the path to fixed image, the reference cine MRI dynamic. The is the path to a cine-MR dynamic of the same imaging session. is the output directory.

Known issues

N/A

Published in

Terpstra ML, Maspero M, D'Agata F, Stemkens B, Intven MP, Lagendijk JJ, Van den Berg CA, Tijssen RH. Deep learning-based image reconstruction and motion estimation from undersampled radial k-space for real-time MRI-guided radiotherapy. Physics in Medicine & Biology. 2020 July 30; 66(15). doi:https://doi.org/10.1088/1361-6560/ab9358.