Preoperative contrast-enhanced computed tomography-based radiomics model for overall survival prediction in hepatocellular carcinoma

Abstract

BACKGROUND Hepatocellular carcinoma (HCC) is the most common primary liver malignancy with a rising incidence worldwide. The prognosis of HCC patients after radical resection remains poor. Radiomics is a novel machine learning method that extracts quantitative features from medical images and provides predictive information of cancer, which can assist with cancer diagnosis, therapeutic decision-making and prognosis improvement. AIM To develop and validate a contrast-enhanced computed tomography-based radiomics model for predicting the overall survival (OS) of HCC patients after radical hepatectomy. METHODS A total of 150 HCC patients were randomly divided into a training cohort ( n = 107) and a validation cohort ( n = 43). Radiomics features were extracted from the entire tumour lesion. The least absolute shrinkage and selection operator algorithm was applied for the selection of radiomics features and the construction of the radiomics signature. Univariate and multivariate Cox regression analyses were used to identify the independent prognostic factors and develop the predictive nomogram, incorporating clinicopathological characteristics and the radiomics signature. The accuracy of the nomogram was assessed with the concordance index, receiver operating characteristic (ROC) curve and calibration curve. The clinical utility was evaluated by decision curve analysis (DCA). Kaplan–Meier methodology was used to compare the survival between the low- and high-risk subgroups. RESULTS In total, seven radiomics features were selected to construct the radiomics signature. According to the results of univariate and multivariate Cox regression analyses, alpha-fetoprotein (AFP), neutrophil-to-lymphocyte ratio (NLR) and radiomics signature were included to build the nomogram. The C-indices of the nomogram in the training and validation cohorts were 0.736 and 0.774, respectively. ROC curve analysis for predicting 1-, 3-, and 5-year OS confirmed satisfactory accuracy [training cohort, area under the curve (AUC) = 0.850, 0.791 and 0.823, respectively; validation cohort, AUC = 0.905, 0.884 and 0.911, respectively]. The calibration curve analysis indicated a good agreement between the nomogram-prediction and actual survival. DCA curves suggested that the nomogram had more benefit than traditional staging system models. Kaplan–Meier survival analysis indicated that patients in the low-risk group had longer OS and disease-free survival (all P < 0.0001). CONCLUSION The nomogram containing the radiomics signature, NLR and AFP is a reliable tool for predicting the OS of HCC patients.

Deep learning model based on contrast-enhanced MRI for predicting post-surgical survival in patients with hepatocellular carcinoma

“…Liu et al [ 28 ] developed a DL-radiomics model using preoperative contrast-enhanced ultrasound images for early-stage HCC to predict progression-free survival after ablation or surgical resection. And some scholars used the radiomics model based on single-center CT images to predict OS of HCC patients after surgery [ 29 , 30 ]. The performance of these models was similar to that of our study, while the generalization performance of the model needs to be further validated with limitations of the single-center study.…”

Computed tomography radiomics signature via machine learning predicts RRM2 and overall survival in hepatocellular carcinoma

“…Prognostic models based on radiomic features combined with clinicopathological features have been recognized by an increasing number of studies. For example, Deng et al ( 25 ) predicted the OS of individuals with HCC following radical hepatectomy by incorporating AFP, neutrophil-to-lymphocyte ratio (NLR), and radiomic features into a nomogram. The AUC of the ROC curve for 1-, 3-, and 5-year OS prediction was 0.850, 0.791, and 0.823 in the training cohort and 0.905, 0.884, and 0.911 in the validation cohort, respectively.…”

Machine Learning Combined with Radiomics Facilitating the Personal Treatment of Malignant Liver Tumors

“…Deng et al [43] demonstrated that radiomics models hold an advantage in predicting overall survival (OS) following radical resection. By constructing a radiomics prediction model using pre-surgery CT images, they achieved AUCs of 0.905, 0.884, and 0.911 for predicting 1-year, 3-year, and 5-year OS, respectively, in the validation cohort.…”