Use of a Noise Optimized Monoenergetic Algorithm for Patient-Size Independent Selection of an Optimal Energy Level During Dual-Energy CT of the Pancreas

Purpose: To investigate the impact of a second-generation noise-optimized monoenergetic algorithm on selection of the optimal energy level, image quality, and effect of patient body habitus for dual-energy multidetector computed tomography of the pancreas. Materials and Methods: Fifty-nine patients (38men, 21 women) underwent dual-energy multidetector computed tomography (80/ Sn140 kV) in the pancreatic parenchymal phase. Image data sets, at energy levels ranging from 40 to 80 keV (in 5-keV increments), were reconstructed using first-generation and second-generation noise-optimized monoenergetic algorithm. Noise, pancreatic contrast-to-noise ratio (CNRpancreas), and CNR with a noise constraint (CNRNC) were calculated and compared among the different reconstructed data sets. Qualitative assessment of image quality was performed by 3 readers. Results: For all energy levels below 70 keV, noise was significantly lower (P <= 0.05) and CNRpancreas significantly higher (P < 0.001), with the second-generation monoenergetic algorithm. Furthermore, the second-generation algorithm was less susceptible to variability related to patient body habitus in the selection of the optimal energy level. The maximal CNRpancreas occurred at 40 keV in 98% (58 of 59) of patients with the second-generation monoenergetic algorithm. However, the CNRNC and readers' image quality scores showed that, even with a second-generation monoenergetic algorithm, higher reconstructed energy levels (60-65 keV) represented the optimal energy level. Conclusions: Second-generation noise-optimized monoenergetic algorithm can improve the image quality of lower-energy monoenergetic images of the pancreas, while decreasing the variability related to patient body habitus in selection of the optimal energy level.