Published online Apr 7, 2023. doi: 10.3748/wjg.v29.i13.2001
Peer-review started: December 12, 2022
First decision: January 22, 2023
Revised: February 1, 2023
Accepted: March 20, 2023
Article in press: March 20, 2023
Published online: April 7, 2023
Processing time: 116 Days and 2.9 Hours
Macrotrabecular-massive hepatocellular carcinoma (MTM-HCC) shows an aggressive phenotype. Early diagnosis of MTM-HCC is beneficial to prevent early recurrence and improve prognosis. Radiomics can convert medical images into high-throughput quantification features, which greatly push the development of precision medicine.
Currently, magnetic resonance imaging (MRI) features have been successfully applied to identify MTM-HCC but have mainly focused on the qualitative analysis of imaging features. In this study, we systematically analysed radiomics, clinical and radiological features to build a more comprehensive prediction model. We aimed to develop a noninvasive model for the preoperative prediction of MTM-HCC.
In this study, we aimed to establish and verify a nomogram based on contrast-enhanced MRI for preoperatively identifying MTM-HCC by comparing different machine learning algorithms.
A total of 232 (training set, 162; test set, 70) hepatocellular carcinoma patients were enrolled. Radiomics features were extracted from contrast-enhanced MRI, followed by dimension reduction. Logistic regression (LR), K-nearest neighbour, Bayes, Tree, and support vector machine algorithms were used to construct radiomics signatures. The relative standard deviation (RSD) was used to quantify the stability of these five algorithms. Multivariable logistic analysis was used to select the useful clinical and radiological features, and different predictive models were established. The performances of the different models were assessed using the area under the curve (AUC).
The LR algorithm with the smaller RSD (3.8%) was used to construct the best radiomics signature, which performed well with AUCs of 0.766 and 0.739 in the training and test sets, respectively. Age, alpha-fetoprotein, tumour size, tumour-to-liver apparent diffusion coefficient ratio, and radiomics score were identified as independent predictors of MTM-HCC to build the nomogram, which performed best with AUCs of 0.896 and 0.805 in the training and test sets, respectively.
The nomogram is a reliable tool for preoperatively identifying the MTM-HCC subtype.
More precise and reliable tools are urgently needed to predict the MTM-HCC subtype. Radiomics is a new method to convert medical images into high-throughput quantification features. In this study, we aimed to develop a dynamic contrast-enhanced MRI-based nomogram for preoperatively identifying the MTM-HCC subtype.