Assessment of the Hepatic Veins in Poor Contrast Conditions using Dual Energy CT: Evaluation of a Novel Monoenergetic Extrapolation Software Algorithm

Purpose: To evaluate a novel monoenergetic post-processing algorithm (MEI+) in patients with poor intrahepatic contrast enhancement. Materials and Methods: 25 patients were retrospectively included in this study. Late-phase imaging of the upper abdomen, which was acquired in dual-energy mode (100/140 kV), was used as a model for poor intrahepatic contrast enhancement. Traditional monoenergetic images (MEI), linearly weighted mixed images with different mixing ratios (MI), sole 100 and 140 kV and MEI+ images were calculated. MEI+ is a novel technique which applies frequency-based mixing of the low keV images and an image of optimal keV from a noise perspective to combine the benefits of both image stacks. The signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of the intrahepatic vasculature (IHV) and liver parenchyma (LP) were objectively measured and depiction of IHV was subjectively rated and correlated with portal venous imaging by two readers in consensus. Results: MEI+ was able to increase the SNR of the IHV (5.7 +/- 0.4 at 40keV) and LP (4.9 +/- 1.0 at 90keV) and CNR (2.1 +/- 0.6 at 40keV) greatly compared to MEI (5.1 +/- 1.1 at 80keV, 4.7 +/- 1.0 at 80keV, 1.0 +/- 0.4 at 70keV), MI (5.2 +/- 1.1 M5:5, 4.8 +/- 1.0 M5: 5, 1.0 +/- 3.5 M9:1), sole 100 kV images (4.4 +/- 1.0, 3.7 +/- 0.8, 1.0 +/- 0.3) and 140 kV images (2.8 +/- 0.5, 3.1 +/- 0.6, 0.1 +/- 0.2). Subjective assessment rated MEI+ of virtual 40 keV superior to all other images. Conclusion: MEI+ is a very promising algorithm for monoenergetic extrapolation which is able to overcome noise limitations associated with traditional monoenergetic techniques at low virtual keV levels and consequently does not suffer from a decline of SNR and CNR at low keV values. This algorithm allows an improvement of IHV depiction in the presence of poor contrast. Key points: The evaluated new image-based algorithm for virtual monoenergetic imaging allows calculating low virtual keV images from dual energy datasets with significantly improved contrast-to-noise ratios. The image based novel monoenergetic extrapolation algorithm applies frequency-based mixing of the low keV images and an image of optimal keV from a noise perspective to combine the benefits of both image stacks. When compared to traditional monoenergetic images, the novel monoenergetic algorithm has improved contrast-to-noise ratios for both low and high virtual keV images. Contrast-enhanced dual energy images with poor contrast conditions can be significantly improved, e. g. late phase imaging of the liver.