Determination of CT Texture Variability Among Several CT Scanners
E Christensen*, L Hunter, F Stingo, S Klawikowski, L Court, UT MD Anderson Cancer Center, Houston, TXTU-C-103-8 Tuesday 10:30AM - 12:30PM Room: 103
Purpose: Radiomics is an emerging field with the potential to use CT textures to predict clinical outcomes for cancer patients. The effect of different scanner use on CT texture variability is poorly understood. The purpose of this study was to compare the effect of scanner used on CT texture variability.
Methods: A thoracic anthropomorphic phantom (Kyoto Kagaku) containing nodules mimicking lung tumors was scanned using four different CT scanners (2 GE, 2 Philips) at nearly identical settings. Nine different regions of interest (ROIs) were identified: 5 lung nodules, 3 bronchial tree regions, and the subcarinal nodal area. Each ROI was contoured using automated contouring parameters in the Pinnacle treatment planning system. Code developed and validated previously was used to extract CT texture features for each ROI. Previous work by our group identified 20 CT texture tumor features that are reproducible (CTr) between repeat scans among patients with lung cancer. These 20 CTr textures were extracted from each of the phantom ROIs. Phantom ROIs with 75% or more of the extracted texture features falling within the patient range were pooled for each machine. A one-way ANOVA was conducted to compare the effect of CT scanner used on 20 CTr textures from pooled ROIs.
Results: Five (55.5%) of the nine phantom ROIs had 75.0-95.7% of CTr textures within the range of patient CTr textures. There was no significant effect of scanner used on any of the 20 CTr textures assessed (p > 0.18).
Conclusion: Using different CT scanners at nearly identical settings does not significantly contribute to variance in CTr texture output. This phantom provides a tool with which institutions can evaluate their scanners for CT texture variances. Future studies to determine the contribution of differences in CT scanner settings, if any, to variances in CT texture will be important.