Skip to main content
. 2025 Dec 8;16:273. doi: 10.1186/s13244-025-02150-y

Table 2.

Qualitative and quantitative analysis of DWI image quality with and without DLR

Qualitative analysis (n = 193) DWI without DLR DWI with DLR p-value
 Artifacts 3.84 ± 0.99 3.61 ± 0.82 0.08
 Lesion conspicuity 3.73 ± 0.95 4.13 ± 0.97 < 0.001
 Liver edge sharpness and vessel clarity 3.51 ± 0.98 3.84 ± 0.99 < 0.001
 Overall image quality 3.68 ± 0.93 4.01 ± 0.94 < 0.001
Quantitative analysis (n = 182)
 SNRLiver 36.3 ± 16.1 39.6 ± 14.3 0.039
 SNRLesion 56.9 ± 28.6 63.5 ± 26.3 0.006
 CNR 33.8 ± 27.6 39.7 ± 24.5 0.006
 ERD (mm) 3.34 ± 0.39 2.38 ± 0.36 < 0.001

Data are shown as mean values  ±  deviation

DLR deep learning reconstruction, SNRLiver signal-to-noise ratio of liver, SNRLesion signal-to-noise ratio of lesion, CNR contrast-to-noise ratio, ERD edge rise distance