Virtual Monoenergetic Dual-Energy Computed Tomography

Objectives The aim of this study was to evaluate the effects on objective and subjective image quality of virtual monoenergetic reconstructions at various energy levels of dual-energy computed tomography (DECT) in patients with head and neck cancer. Materials and Methods We included 71 (53 men, 18 women; age, 59.3 ± 12.0 years; range, 33–90 years) patients with biopsy-proven untreated primary (n = 55) or recurrent (n = 16) squamous cell carcinoma who underwent head and neck DECT. Images were reconstructed with a linear blending setting emulating 120 kV acquisition (M_0.3; 30% of 80 kV, 70% of 140 kV spectrum) and as virtual monoenergetic images with photon energies of 40, 60, 80, and 100 keV. Attenuation of lesion, various anatomic landmarks, and image noise were objectively measured, and lesion contrast-to-noise ratio (CNR) was calculated. Two independent blinded radiologists subjectively rated each image series using a 5-point grading scale regarding overall image quality, lesion delineation, image sharpness, and image noise. Results Tumor attenuation peaked at 40 keV (140.2 ± 42.6 HU) followed by the 60 keV (121.7 ± 25.5 HU) and M_0.3 series (102.7 ± 22.3; all P < 0.001). However, the calculated lesion CNR was highest in the 60 keV reconstructions (12.45 ± 7.17), 80 keV reconstructions (8.66 ± 6.58), and M_0.3 series (5.21 ± 3.15; all P < 0.001) and superior to the other monoenergetic series (all P < 0.001). Subjective image analysis was highest for the 60 keV series regarding overall image quality (4.22; κ = 0.411) and lesion delineation (4.35; κ = 0.459) followed by the M_0.3 series (3.81; κ = 0.394; 3.77; κ = 0.451; all P < 0.001). Image sharpness showed no significant difference between both series (3.81 vs 3.79; P = 0.78). Image noise was rated superior in the 80 and 100 keV series (4.31 vs 4.34; P = 0.522). Conclusions Compared with linearly blended images, virtual monoenergetic reconstructions of DECT data at 60 keV significantly improve lesion enhancement and CNR, subjective overall image quality, and tumor delineation of head and neck squamous cell carcinoma.