Evaluation of a second-generation monoenergetic reconstruction algorithm for lesion contrast and venous invasion in pancreatic ductal adenocarcinomas.

Abstract

At lower energy levels, virtual monochromatic imaging by dual-energy computed tomography improves lesion attenuation but produces greater image noise with the conventional monoenergetic reconstruction algorithm (Mono). Recently, a second-generation algorithm (Mono+) was introduced to overcome this limitation. We compared the quality of images obtained with these algorithms and investigated the optimal energy selection for pancreatic ductal adenocarcinomas (PDACs). Image data from 54 PDAC cases were generated at 40, 50, 60, 70, and 80 keV using Mono and Mono+. Image quality was objectively assessed by comparing the signal-to-noise ratios (SNRs), noise, and the contrast-to-noise ratios (CNRs) at different keV levels and between these algorithms at the same keV level. Lesion conspicuity and venous invasion were subjectively assessed. For Mono, the mean pancreas and tumour SNRs peaked at 70 keV (p<0.001). The noise increased as the energy level decreased (p<0.001). CNRtumour remained unchanged. For Mono+, the mean pancreas SNR peaked at 40 keV (p<0.001). The mean tumour SNR and noise remained unchanged. The tumour CNRs were highest at 40 keV (4.9 times the CNR of Mono 40 keV, p<0.001). Subjectively, lesion conspicuity was best at Mono+ 40 keV (p<0.001) and it showed higher diagnostic performance levels on venous invasion assessment against Mono. Mono+ produced better image quality, and 40 keV is recommended for the diagnosis of PDAC.