Dual-energy computed tomography in patients with cutaneous malignant melanoma: Comparison of noise-optimized and traditional virtual monoenergetic imaging

Objective: The aim of this study was to investigate the impact of noise-optimized virtual monoenergetic imaging (VMI +) reconstructions on quantitative and qualitative image parameters in patients with cutaneous malignant melanoma at thoracoabdominal dual-energy computed tomography (DECT). Materials and methods: Seventy-six patients (48 men; 66.6 +/- 13.8 years) with metastatic cutaneous malignant melanoma underwent DECT of the thorax and abdomen. Images were post-processed with standard linear blending (M_0.6), traditional virtual monoenergetic (VMI), and VMI+ technique. VMI and VMI + images were reconstructed in 10-keV intervals from 40 to 100 keV. Attenuation measurements were performed in cutaneous melanoma lesions, as well as in regional lymph node, subcutaneous and in-transit metastases to calculate objective signal-to-noise (SNR) and contrast-to-noise (CNR) ratios. Five-point scales were used to evaluate overall image quality and lesion delineation by three radiologists with different levels of experience. Results: Objective indices SNR and CNR were highest at 40-keV VMI + series (5.6 +/- 2.6 and 12.4 +/- 3.4), significantly superior to all other reconstructions (all P < 0.001). Qualitative image parameters showed highest values for 50-keV and 60-keV VMI + reconstructions (median 5, respectively; P <= 0.019) regarding overall image quality. Moreover, qualitative assessment of lesion delineation peaked in 40-keV VMI + (median 5) and 50-keV VMI + (median 4; P = 0.055), significantly superior to all other reconstructions (all P < 0.001). Conclusion: Low-keV noise-optimized VMI + reconstructions substantially increase quantitative and qualitative image parameters, as well as subjective lesion delineation compared to standard image reconstruction and traditional VMI in patients with cutaneous malignant melanoma at thoracoabdominal DECT.