Skip to main content
. 2022 Feb 21;12(2):550. doi: 10.3390/diagnostics12020550

Figure 2.

Figure 2

Overall algorithm scheme. First, the computed tomography (CT) volume was preprocessed; then, the liver and spleen were segmented as the region of interest. The radiomic features were extracted from the segmented liver and spleen. The machine learning classifiers were trained by hepatic features and hepatic–splenic features, respectively, to predict the probability array of liver fibrosis stages, namely F0–F4.