SU-E-CAMPUS-J-02: Exploring Appropriate CT Acquisition Parameters for Measuring Tumor Volumes in Response Assessment Using An Anthropomorphic Thorax Phantom

Abstract

Purpose: Early tumor volume change holds promise for detecting the biologic activity of systemic therapy in non‐small cell lung cancer. It is imperative to optimize and standardize CT imaging protocols to further investigate volumetric response. This study explored the effects of CT acquisition parameters on volume measurements using a thorax phantom and segmentation algorithm. Methods: Eight phantom nodules of known volume (small size of 520–540mm^3, nominal diameter of 1mm and large size of 4210–4320mm^3, nominal diameter of 2mm), shape (spherical, elliptical) and density (−10 and +100 HU) were scanned on a GE VCT at four doses (25, 50, 100, and 200 mAs). For each scan, six image series were reconstructed at three slice intervals of 5, 2.5 and 1.25mm, and using the lung and standard reconstruction algorithms. Each nodule's volume was measured using an in‐house segmentation algorithm. Relative difference in volume (RDiV = (Algorithm‐True)/True*100%) was calculated. Linear mixed effects models were fitted to examine the effect of dose, and the combinations of slice thickness and reconstruction algorithm. Results: Dose did not affect the RDiV after adjusting for slice thickness and reconstruction algorithm (p=0.40). The combination of 1.25mm/standard had an estimated RDiv of −0.001 which was closest to 0. Slice thickness of 1.25mm/lung and 2.5mm/standard were not statistically different from 1.25mm/standard. Estimated differences of −6.3%(p=0.002), 7.1% (p<0.001), and 7.3%(p<0.001) in RDiV were observed for 2.5mm/lung, 5mm/standard and 5mm/lung when compared to 1.25mm/standard, respectively. Conclusion: Imaging acquisition parameters of CT slice thickness and reconstruction algorithm affect the accuracy and precision of lung nodule volume measurements to various degrees while dose has almost no effect. Our findings may help determine appropriate CT imaging acquisition parameters for clinical trials and care. Validation studies using different segmentation algorithms are warranted. This work is in part supported by Dr. Binsheng Zhao's NIH R01 CA149490. No other conflict of interest.