Room: Track 1
CT number accuracy action limits for the ACR CT phantom are a widely-accepted standard in CT quality control (QC). Spectral CT imaging, however, has no well-established action limits for QC. This work explores establishing action limits for spectral CT QC using longitudinal measurements and ACR action limits for conventional CTs.
An ACR CT phantom was scanned weekly on a dual-layer-detector spectral CT scanner (iQon, Philips). Conventional and spectral images were obtained simultaneously. Conventional, monoenergetic (40, 70 and 200 keV), iodine-no-water and virtual noncontrast (VNC) images from 57 continuous weekly 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= week 1 … week 57) was used for analysis. For each insert and image type, the longitudinal standard deviation (SD) of CT_ROI were calculated. The SD ratios between the spectral and conventional CT measurements were calculated and used as conversion factors to obtain action limits for spectral CT. Water CT number action limit of [-5, 5] HU from conventional CT was used as the baseline.
The conversion factor from conventional CT to iodine-no-water images was 0.089 (0.11-0.065, 95% confident interval, unit: mg/ml per HU), which gave ±1.73 mg/ml action limit for the bone insert in the ACR phantom. The conversion factors for VNC and monoenergetic images were close to 1, except 3.76 (4.43-3.10) for water at 40 keV and 6.57 (8.37-4.77) for bone at 40 keV, which means action limits of ±18.7 HU and ±125.7 HU, respectively.
We established action limits of spectral measurements, which can be used directly in spectral CT QC. When measurements exceed the limits, action needs to be taken to correct the issue.