Impact of motion velocity on four-dimensional target volumes: A phantom study

This study aims to assess the impact of motion velocity that may cause motion artifacts on target volumes (TVs) using a one-dimensional moving phantom. A 20 mm diameter spherical object embedded in a QUASAR (TM) phantom sinusoidally moved with approximately 5.0 or 10.0 mm amplitude (A) along the longitudinal axis of the computed tomography (CT) couch. The motion period was manually set in the range of 2.0-10.0 s at approximately 2.0 s interval. Four-dimensional (4D) CT images were acquired by a four-slice CT scanner (LightSpeed RT; General Electric Medical Systems, Waukesha, WI) with a slice thickness of 1.25 mm in axial cine mode. The minimum gantry rotation of 1.0 s was employed to achieve the maximum in-slice temporal resolution. Projection data over a full gantry rotation (1.0 s) were used for image reconstruction. Reflective marker position was recorded by the real-time positioning management system (Varian Medical Systems, Palo Alto, CA). ADVANTAGE 4D software exported ten respiratory phase volumes and the maximum intensity volume generated from all reconstructed data (MIV). The threshold to obtain static object volume (V-0, 4.19 ml) was used to automatically segment TVs on CT images, and then the union of TVs on 4D CT images (TV4D) was constructed. TVs on MIV (TVMIV) were also segmented by the threshold that can determine the area occupied within the central slice of TVMIV. The maximum motion velocity for each phase bin was calculated using the actual averaged motion period displayed on ADVANTAGE 4D software (T), the range of phases used to construct the target phase bin (phase range), and a mathematical model of sinusoidal function. Each volume size and the motion range of TV in the cranial-caudal (CC) direction were measured. Subsequently, cross-correlation coefficients between TV size and motion velocity as well as phase range were calculated. Both misalignment and motion-blurring artifacts were caused by high motion velocity. Less than 6% phase range was needed to construct the 4D CT data set, except for T of 2.0 s. While the positional differences between the TV and ideal centroid in the CC direction were within the voxel size for T >= 6.0 s, the differences were up to 2.43 and 4.15 mm for (A,T)=(5.0 mm,2.0 s) and (10.0 mm, 2.0 s), respectively. The maximum volumetric deviations between TV sizes and V-0 were 43.68% and 91.41% for A of 5.0 and 10.0 mm, respectively. TVMIV sizes were slightly larger than TV4D sizes. Volumetric deviation between TV size and V-0 had a stronger correlation with motion velocity rather than phase range. This phantom study demonstrated that motion artifacts were substantially reduced when the phantom moved longitudinally at low motion velocity during 4D CT image acquisition; therefore, geometrical uncertainties due to motion artifacts should be recognized when determining TVs, especially with a fast period.