SU‐F‐T‐253: Volumetric Comparison Between 4D CT Amplitude and Phase Binning Mode

Abstract

Purpose:Motion artifact in 4DCT images can affect radiation treatment quality. To identify the most robust and accurate binning method, we compare the volume difference between targets delineated on amplitude and phase binned 4DCT scans.Methods:Varian RPM system and CT scanner were used to acquire 4DCTs of a Quasar phantom with embedded cubic and spherical objects having superior‐inferior motion. Eight patients’ respiration waveforms were used to drive the phantom. The 4DCT scan was reconstructed into 10 phase and 10 amplitude bins (2 mm slices). A scan of the static phantom was also acquired. For each waveform, sphere and cube volumes were generated automatically on each phase using HU thresholding. Phase (amplitude) ITVs were the union of object volumes over all phase (amplitude) binned images. The sphere and cube volumes measured in the static phantom scan were Vsphere=4.19cc and Vcube=27.0cc. Volume difference (VD) and dice similarity coefficient (DSC) of the ITVs, and mean volume error (MVE) defined as the average target volume percentage difference between each phase image and the static image, were used to evaluate the performance of amplitude and phase binning.Results:Averaged over the eight breathing traces, the VD and DSC of the internal target volume (ITV) between amplitude and phase binning were 3.4%±3.2% (mean ± std) and 95.9%±2.1% for sphere; 2.1%±3.3% and 98.0% ±1.5% for cube, respectively.For all waveforms, the average sphere MVE of amplitude and phase binning was 6.5% ± 5.0% and 8.2%±6.3%, respectively; and the average cube MVE of amplitude and phase binning was 5.7%±3.5%and 12.9%±8.9%, respectively.Conclusion:ITV volume and spatial overlap as assessed by VD and DSC are similar between amplitude and phase binning. Compared to phase binning, amplitude binning results in lower MVE suggesting it is less susceptible to motion artifact.