MRI-based radiomics model and nomogram for predicting the outcome of locoregional treatment in patients with hepatocellular carcinoma

BMC Medical Imaging

Research

Zhenhao Liu¹  Yuxin Wang²  Jiahui Jiang²  Himeko Asayo²  Hui Xu²  Dawei Yang²  Zhenghan Yang² 

[1] Department of Radiology, Affiliated Hospital of Changzhi Institute of Traditional Chinese Medicine, 046099, Changzhi, China;Department of Radiology, Beijing Friendship Hospital, Capital Medical University, 100050, Beijing, China;

关键词: Hepatocellular carcinoma;    Radiomics;    Magnetic resonance imaging;    Locoregional treatment;   

DOI  :  10.1186/s12880-023-01030-5

 received in 2023-04-02, accepted in 2023-05-23,  发布年份 2023

来源: Springer

【 摘 要 】

BackgroundPrediction of locoregional treatment response is important for further therapeutic strategy in patients with hepatocellular carcinoma. This study aimed to investigate the role of MRI-based radiomics and nomogram for predicting the outcome of locoregional treatment in patients with hepatocellular carcinoma.MethodsThe initial postoperative MRI after locoregional treatment in 100 patients with hepatocellular carcinoma was retrospectively analysed. The outcome was evaluated according to mRECIST at 6 months. We delineated the tumour volume of interest on arterial phase, portal venous phase and T2WI. The radiomics features were selected by using the independent sample t test or nonparametric Mann‒Whitney U test and the least absolute shrinkage and selection operator. The clinical variables were selected by using univariate analysis and multivariate analysis. The radiomics model and combined model were constructed via multivariate logistic regression analysis. A nomogram was constructed that incorporated the Rad score and selected clinical variables.ResultsFifty patients had an objective response, and fifty patients had a nonresponse. Nine radiomics features in the arterial phase were selected, but none of the portal venous phase or T2WI radiomics features were predictive of the treatment response. The best radiomics model showed an AUC of 0.833. Two clinical variables (hCRP and therapy method) were selected. The AUC of the combined model was 0.867. There was no significant difference in the AUC between the combined model and the best radiomics model (P = 0.573). Decision curve analysis demonstrated the nomogram has satisfactory predictive value.ConclusionsMRI-based radiomics analysis may serve as a promising and noninvasive tool to predict outcome of locoregional treatment in HCC patients, which will facilitate the individualized follow-up and further therapeutic strategies guidance.

【 授权许可】

CC BY   
© The Author(s) 2023

【 图 表 】

41116_2023_36_Article_IEq145.gif

Fig. 5

41116_2023_36_Article_IEq134.gif

41116_2023_36_Article_IEq131.gif