Noise-optimized advanced image-based virtual monoenergetic imaging for improved visualization of lung cancer: Comparison with traditional virtual monoenergetic imaging

Abstract

PurposeTo assess the effect of a noise-optimized image-based virtual monoenergetic imaging (VMI+) algorithm in direct comparison with the traditional VMI technique and standard linearly-blended images emulating 120-kVp acquisition (M_0.3) on image quality at dual-energy CT in patients with lung cancer. Materials and MethodsDual-source dual-energy CT examinations of 48 patients with biopsy-proven primary (n=31) or recurrent (n=20) lung cancer were evaluated. Images were reconstructed as M_0.3, and VMI+ and traditional VMI series at 40, 55, and 70keV. Attenuation of tumor, descending aorta, pulmonary trunk, latissimus muscle, and noise were measured. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were calculated. Five-point scales were used by three observers to subjectively evaluate general image impression, tumor delineation, image sharpness, and image noise. ResultsBackground noise was consistently lower with VMI+ compared to VMI at all keV levels (all p<0.0001) and M_0.3 (all p≤0.0004). Tumor SNR and CNR peaked in the 40keV VMI+ series, significantly higher compared to all VMI and M_0.3 series (all p<0.0008). Observers preferred the 55keV VMI+ series regarding general image impression and tumor delineation compared to all other series (all p<0.0001). Image sharpness and image noise ratings were highest in the 55keV VMI+ and 70keV VMI and VMI+ reconstructions. ConclusionsTumor CNR peaked at 40keV VMI+ while observers preferred 55keV VMI+ series overall other series for dual-energy CT of lung cancer. The noise-optimized VMI+ technique showed significantly lower background noise and higher SNR and CNR compared to the traditional VMI technique at matching keV levels.