Virtual non-contrast images from dual-layer spectral CT: comparison with true non-contrast across abdominal structures

OBJECTIVE: To compare virtual non-contrast (VNC) images derived from arterial (VNC_A) and portal venous phase (VNC_P) of dual-layer spectral CT (DLCT) with true non-contrast (TNC) images in abdominal imaging, assessing attenuation accuracy, image quality, lesion and calcification measurements, and the potential of VNC to replace TNC in routine protocols. MATERIALS AND METHODS: One hundred consecutive patients undergoing triphasic abdominal CT were included. TNC, arterial, and portal venous phases were acquired on a DLCT scanner. VNC_A and VNC_P images were generated using a two-material decomposition algorithm. Attenuation values, image noise, signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and lesion and calcification metrics were measured. Subjective image quality and diagnostic acceptability were graded by three radiologists. Agreement with TNC was evaluated using ICC, Bland-Altman analysis, ANOVA, and non-parametric tests. Radiation dose reduction achievable by omitting the TNC phase was also assessed. RESULTS: Although VNC_A showed better agreement with TNC than VNC_P, both consistently underestimated parenchymal and lesion attenuation; lesion size was preserved (12.2 ± 7.9 mm; p = 0.996). Calcification density was significantly reduced in both VNC phases, and calcification diameter was slightly underestimated in VNC_A (p < 0.001). VNC images demonstrated higher or similar SNR and CNR but inferior subjective image quality (p < 0.001), due to blotchier appearance. Only 12% of VNC_P datasets were judged fully interchangeable with TNC. Replacing TNC with VNC would yield a 33% radiation dose reduction in this specific protocol. CONCLUSIONS: VNC cannot replace TNC based on the current evidence; further refinements are required for broader clinical adoption. KEY POINTS: Question True non-contrast images (TNC) are essential in abdominal CT, yet it remains unclear whether spectral CT-derived virtual non-contrast images (VNC) can reliably replace TNC imaging. Findings VNC images approximate TNC images but introduce consistent attenuation biases, reduced subjective quality, and altered calcification metrics, preventing full replacement of dedicated non-contrast acquisitions. Clinical relevance Replacing TNC with VNC images reduces radiation exposure, yet current VNC performance shows measurable differences from TNC, indicating that algorithmic refinement is required before routine diagnostic use in abdominal CT.