Room: Exhibit Hall | Forum 9
Purpose: To evaluate the longitudinal variation and correlation of spectral CT measurements using daily QC scan of an ACR CT phantom.
Methods: An ACR CT phantom was scanned daily on a dual-layer detector, spectral CT scanner (iQon, Philips) as part of a CT QC program. Conventional and spectral images were obtained simultaneously. Conventional, monoenergetic (40, 70 and 200 keV), iodine-no-water (mg iodine/ml) and virtual noncontrast (VNC) images from 30 continuous daily scans were analyzed. The central slice of the CT number accuracy module was used. Regions of interest (ROI) were drawn inside each of the four inserts (water, bone, polyethylene, and acrylic) and the mean value for each ROI (CT_ROI(i), i= day 1 â€¦ day 30) was used for analysis. For each insert, longitudinal standard deviation (SD) and Pearson correlation of CT_ROI were calculated. Relative SD of conventional, monoenergetic and VNC images were calculated using SD(CT_ROI)/(mean(CT_ROI) + 1000). The relative SD in the iodine-no-water image was calculated using SD(CT_ROI)/mean(CT_ROI) to enable direct comparison with the other results.
Results: In conventional imaging, the longitudinal variations were 0.09% (water), 0.22% (bone), 0.11% (polyethylene) and 0.12% (acrylic). In spectral imaging, the water insert variation ranged from 0.29% (40 keV) to 0.06% (200 keV). Higher variation was observed in the bone insert, e.g., 1.86% in iodine-no-water image and 0.90% in 40 keV image. There were strong correlations (r>0.86, p <0.05) between most spectral results for the same insert, except the measurements for the 200 keV image had weak correlation with other spectral images for water, polyethylene and acrylic (r<0.44).
Conclusion: Higher longitudinal variation may make quantitative comparison using spectral CT imaging over time challenging. However, due to strong correlation of CT measurements across spectral image types, routine consistency checks for all the image types may not be necessary.