Rationale and objective: To develop and evaluate a method for predicting and reducing motion artifacts in free-breathing liver perfusion computed tomography (CT) scanning with couch shuttling and to compare tumor and liver parenchyma perfusion values.
Materials and methods: Thirty patients (23 males, 7 females, median age of 74 years) with primary or metastatic intrahepatic tumors underwent dynamic contrast enhanced CT scans with axial shuttling. A semiautomatic respiratory motion correction algorithm was applied to align the acquired images along the z-axis. Perfusion maps were generated using the dual-input Johnson-Wilson model. Root mean squared deviation (RMSD) maps of the model fit to the pixel time-density curves were calculated.
Results: Precorrection RMSD correlated positively with magnitude of change in functional values resulting from motion. Blood flow, arterial blood flow, and permeability surface product were significantly increased in tumor compared to normal tissue (P < .05), blood volume was significantly reduced in tumor compared to normal tissue (P < .05). In a subgroup of patients with high-amplitude motion significant difference was observed between uncorrected and motion correction blood flow maps.
Conclusions: Patients can breathe freely during hepatic perfusion imaging if retrospective motion correction is applied to reduce motion artifacts. RMSD provides a regional assessment of motion induced artifacts in liver perfusion maps.
Copyright © 2013 AUR. Published by Elsevier Inc. All rights reserved.