Comparison of flat-panel-detector-based CT and multidetector-row CT in automated volumetry of pulmonary nodules using an anthropomorphic chest phantom

Abstract

This study evaluates the accuracy and reproducibility of an experimental flat-panel-detector-based CT scanner (fp-CT) in comparison with those of a 64-slice multidetector row CT (MDCT) in automated pulmonary nodule volumetry. An anthropomorphic chest phantom with 31 spherical nodules (nodule diameters of 2.94-10.01 mm; volumes of 13.24-524.97 mm(3)) was scanned both with an amorphous silicon-based fp-CT scanner, using various tube current and kilovoltage settings, and with a conventional MDCT scanner. Automated nodule volumetry was performed using dedicated software. CT image data were evaluated twice by two independent radiologists. Intra- and inter-observer variations of volumetric measurements were determined and tested using the Kruskal-Wallis test and analysis of variance (fn-ANOVA). The percentage measurement errors (PME) were calculated and differences tested using Wilcoxon signed ranks and Friedman tests. Intraobserver variation was significantly higher for MDCT than for fp-CT (range: p = 0.043-0.045). The measured nodule volumes were significantly greater on fp-CT than on MDCT scans (p<0.001). The PME was significantly greater in fp-CT than in MDCT scans (PME range, 12.35-13.35% for fp-CT scan protocols and 16.87-19.02% for MDCT scan protocols; p<0.0001). The PME increased significantly with reduction of nodule size, and this increase was significantly higher on MDCT than on fp-CT scans (p = 0.0001). The absolute PME was significantly different for nodules of less than 5 mm in diameter (p = 0.0001-0.0033) than for larger nodules. Flat-panel-detector-based CT has advantages over MDCT in accurately determining the volume of pulmonary nodules below 5 mm in diameter.