Click here to

Session: New and Emerging Technology [Return to Session]

Generating a Phantom with a Motion Model Ground Truth

B Lau*, O Dillon, T Reynolds, R O'Brien, University of Sydney, Sydney, NSW, AU,


SU-H330-IePD-F8-1 (Sunday, 7/10/2022) 3:30 PM - 4:00 PM [Eastern Time (GMT-4)]

Exhibit Hall | Forum 8

Purpose: Modern 4DCBCT reconstruction algorithms such as SMEIR and MCFDK attempt to explicitly model motion as part of the image formation process. These reconstruction algorithms are particularly important for the development of faster scans with reduced radiation such as adaptive 4DCBCT. However, simulation studies that assess performance and test new approaches typically lack an explicit motion ground truth. Therefore, we propose a novel 4D motion XCAT (mXCAT) method to assist the assessment of motion compensated reconstruction algorithms.

Methods: Motion was estimated from a 10 respiratory phase 4D XCAT phantom to generate 9 consistent Deformation Vector Fields (DVFs) and volumes. The 4D mXCAT comprised of 9 consistent volumes and the reference phase 1 4D XCAT volume. Evenly spaced 200 projection 4DCBCT was simulated through the 4D mXCAT and 4D XCAT volumes and then reconstructed using FDK and MCMKB algorithms. Reconstructed image quality was quantified using Structural Similarity Index (SSIM), Root-Mean-Squared Error (RMSE), Signal-to-Noise-Ratio (SNR), Contrast-to-Noise-Ratio (CNR), Tissue Interface Width-Diaphragm (TIS-D) Tissue Interface Width-Tumor (TIS-T). DVFs were quantified by calculating the relative distance and absolute distance between the ground truth DVF and the MCMKB DVF.

Results: The mXCAT MCMKB reconstructions yielded a SSIM 0.99, RMSE 0.0027, SNR 75.1, CNR 13.9, TIW-Diaphragm 1.4 and TIW-Tumor of 1.4 compared to XCAT MCMKB reconstructions with SSIM 0.98, RMSE 0.0028, SNR 34.8, CNR 13, TIW-Diaphragm 1.5 and TIW-Tumor 2.0. There was an average of 0.9% relative distance and 2.7 mm absolute distance between the 4D mXCAT ground truth DVF and the mxCAT MCMKB DVF.

Conclusion: The mXCAT enabled comparisons between motion estimates generated during reconstruction directly against a motion ground truth.

Funding Support, Disclosures, and Conflict of Interest: This study was funded by NHMRC grant 1138899 and partly by a Cancer Australia (Priority-driven Collaborative Cancer Research Scheme) project grant number 1161748.


Cone-beam CT, Phantoms, Reconstruction


IM- Cone Beam CT: Image Reconstruction

Contact Email