Evaluation of photon-counting CT-derived attenuation parameters for noninvasive liver fat assessment

To evaluate the diagnostic performance of multiple photon-counting CT (PCCT)-derived parameters for hepatic fat quantification, using MRI-derived proton density fat fraction (PDFF) as the reference standard.

This prospective study included 59 adult participants who underwent both MRI and PCCT within 2 h. Standardized CT values, Liver-virtual non-contrast (VNC) fat fraction, and liver-to-spleen/fat-derived indices (L-S, L/S, L-F, L + F, L/F) were obtained from non-contrast PCCT. MRI PDFF was derived using a 6-echo q-Dixon sequence. Pearson and Spearman correlation analyses were used to assess the association between PCCT-derived parameters and MRI PDFF. Linear mixed-effects model (LMM) was performed to establish a conversion formula between CT attenuation and MRI PDFF.

All PCCT-derived parameters were significantly correlated with MRI PDFF. Standardized CT values demonstrated the strongest correlation (r = -0.94, p < 0.001), followed by L-S (r = -0.90, p < 0.001), L/F (r = 0.89, p < 0.001), and L/S (r = -0.88, p < 0.001). Liver-VNC fat fraction showed a relatively lower correlation (r = 0.84, p < 0.001). LMM revealed a significant association between standardized CT values and MRI PDFF (β = -0.55, p < 0.001), with the regression equation: MRI PDFF (%) = -0.55 × CT (HU) + 41.

Among various attenuation-based PCCT parameters, the standardized CT value shows the strongest correlation with MRI PDFF and is clinically easy to obtain. It may serve as a promising noninvasive alternative to MRI for hepatic fat quantification. However, due to the limited sample size, these findings should be interpreted with caution.