Noise-optimized virtual monoenergetic dual-energy computed tomography: optimization of kiloelectron volt settings in patients with gastrointestinal stromal tumors

Purpose: The aim of this study was to evaluate the impact of a noise-optimized virtual monoenergetic imaging (VMI+) reconstruction technique on quantitative and qualitative image analysis in patients with gastrointestinal stromal tumors (GISTs) at dual-energy computed tomography (DECT) of the abdomen. Methods: Forty-five DECT datasets of 21 patients (14 men; 63.7 +/- 9.2 years) with GISTs were reconstructed with the standard linearly blended (M_ 0.6) and VMI+ and traditional virtual monoenergetic (VMI) algorithm in 10-keV increments from 40 to 100 keV. Attenuation measurements were performed in GIST lesions and abdominal metastases to calculate objective signal-tonoise (SNR) and contrast-to-noise ratios (CNR). Fivepoint scales were used to evaluate overall image quality, lesion delineation, image sharpness, and image noise. Results: Quantitative image parameters peaked at 40-keV VMI+ series (SNR 27.8 +/- 13.0; CNR 26.3 +/- 12.7), significantly superior to linearly blended (SNR 16.8 +/- 7.3; CNR 13.6 +/- 6.9) and all VMI series (all P < 0.001). Qualitative image parameters were highest for 60-keV VMI+ reconstructions regarding overall image quality and image sharpness (median 5, respectively; P <= 0.023). Qualitative assessment of lesion delineation peaked in 40 and 50-keV VMI+ series (median 5, respectively). Image noise was superior in 90 and 100-keV VMI and VMI+ reconstructions (all medians 5). Conclusions: Low-keV VMI+ reconstructions significantly increase SNR and CNR of GISTs and improve quantitative and qualitative image quality of abdominal DECT datasets compared to traditional VMI and standard linearly blended image series.