Abstract: Objective To explore the application value of multimodal nuclear magnetic resonance (MRI) based imaging features in the differential diagnosis of malignant hepatic tumor and benign rich of blood supply hepatic lesions. Methods A total of 50 cases of hepatic malignant tumor (liver cancer group) and 60 cases of benign rich of blood supply hepatic lesions (liver hypertrophic lesion group) were selected from the Surgical Department of Chongming Hospital Affiliated to Shanghai Health Medical College from January 2023 to June 2024. The image omics features of T2-weighted imaging (T₂WI) and diffusion-weighted imaging (DWI) were extracted, and combined with clinical risk factors. Multiple logistic regression and 5-FCV were used to establish a nomogram model. The model performance was evaluated by receiver operating characteristic (ROC) curve, calibration curve and decision-curve-analysis (DCA) method. Results The male ratio, age, cirrhosis, tumor diameter, alpha fetoprotein (AFP) and vascular endothelial growth gactor (VEGF) levels in the liver cancer group were 70.00%, (65.1±5.0) years old, 40.00%, (5.06±1.02) cm, (100.01±50.06) ng/mL, and (5.01±1.56) pg/mL, respectively, which were higher than those of 48.33%, (50.3±14.5) years old, 5.00%, (3.04±1.06) cm, (5.14±1.02) ng/mL, and (2.02±0.01) pg/mL in the benign liver hypertrophic lesion group. The differences were statistically significant (t/χ²=4.510, 6.862, 5.337, 10.123, 14.710, 14.860, all P<0.05). A multi-factor analysis showed that male, age, history of cirrhosis, tumor diameter, increased AFP and VEGF levels were independent predictors of liver malignancy (95%CI: 1.295~1.570, 0.923~1.146, 1.517~1.748, 1.159~1.405, 1.826~2.269, 2.058~2.536, 95%CI: 1.295~1.570, 0.923~1.146, 1.517~1.748, 1.159~1.405, 1.826~2.269, 2.058~2.536). All P<0.05). The calibration curves of the training set and the test set show that the two are basically identical, and the DCA display model has clinical application value in distinguishing liver malignant tumor from benign rich of blood supply hepatic lesions. Conclusion Multimodal MRI imaging features have important application value in the differential diagnosis of liver malignant tumors from benign hypertrophic liver lesions. The prediction model combined with clinical risk factors can effectively distinguish liver malignant tumors and benign hypertrophic liver lesions, and has a good clinical application prospect.

Key words: Liver malignancy, Benign hyperemia of liver, Multimodal MRI, Image features, Molecular mechanism, Appraised value