Computed Tomography (CT) scans are crucial in medical diagnostics, often requiring both unenhanced and contrast-enhanced images. Unenhanced images serve as a baseline for comparison, helping to detect subtle changes after contrast agent administration. However, acquiring separate unenhanced scans increases radiation exposure and workflow complexity. Virtual non-contrast (VNC) images, derived from spectral detector CT, offer a potential solution by generating unenhanced-like images from contrast-enhanced scans. This raises a critical question: How well do VNC images compare to true unenhanced images?
A recent study rigorously investigated this comparison using a novel dual-layer spectral detector CT scanner. Researchers aimed to determine if the attenuation values in VNC images are comparable to those in actual unenhanced images across various body tissues. The study involved 46 patients undergoing both pre- and post-contrast CT scans, resulting in a substantial dataset of 84 paired unenhanced and VNC images across different scan phases (arterial, portal venous/nephrographic, and urographic). Attenuation measurements were meticulously taken in key organs and tissues including the liver, spleen, kidneys, psoas muscle, abdominal aorta, and subcutaneous fat by expert readers.
The findings revealed a strong agreement between VNC and unenhanced images. In a remarkable 92.6% of measurements, the difference in attenuation was less than 15 Hounsfield Units (HU), and in 75.2%, it was less than 10 HU. Statistical equivalence was confirmed for all tissues except subcutaneous fat when using a threshold of 10 HU difference. Interestingly, the phase of contrast enhancement (arterial, venous, or urographic) did not significantly impact the comparability. However, in subcutaneous fat, VNC images tended to overestimate attenuation values compared to true unenhanced images. Image quality assessment indicated that VNC images were generally rated as good to excellent in arterial and nephrographic phases, but quality decreased in the urographic phase.
In conclusion, this study demonstrates that VNC images from dual-layer spectral detector CT provide attenuation values that are largely comparable to true unenhanced images for most tissues. This suggests VNC imaging holds promise for reducing radiation dose and streamlining CT protocols. However, the discrepancy observed in subcutaneous fat highlights the need for caution, particularly when evaluating lesions containing fat. Further research is warranted to determine if current clinical attenuation thresholds need adjustment when using VNC images, especially in scenarios involving fatty tissues. The comparison between VNC and true unenhanced images is crucial for confident clinical translation of this technology.
