Prediction of Ki-67 expression in hepatocellular carcinoma: a dual-center study based on T2-weighted imaging habitat analysis
Xiaojun Zheng, Lihong Huang, Mengjie Huang, Bin Yu, Shiji Qin, Deyou HuangAbstract
Background
Limited research has applied habitat imaging to evaluate the association between T2-weighted magnetic resonance imaging (T2WI-MRI) features and Ki-67 expression in hepatocellular carcinoma HCC). This study aimed to link T2WI habitat-derived parameters to Ki-67 status and aggressiveness in HCC.
Patients and methods
This dual-center retrospective study, enrolled patients with pathologically confirmed HCC undergoing preoperative MRI (2020–2024). Using Ki-67 index cutoff (20%). Tumor habitat partitioning was performed using k-means clustering (k = 5), followed by extraction of both habitat-specific radiomic features and conventional whole-tumor features. Feature selection was conducted using least absolute shrinkage and selection operator (LASSO) regression with 10-fold cross-validation. Three predictive models were constructed with the ExtraTrees algorithm: a habitat model, a radiomics model, and a clinical model. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), DeLong test, and decision curve analysis (DCA).
Results
T2WI-based habitat imaging enables a noninvasive assessment of intratumoral heterogeneity, significantly improving the prediction of Ki-67 expression status in HCC. This approach may provide a promising imaging biomarker for molecular subtyping and support personalized preoperative treatment strategies.
Conclusions
T2WI habitat imaging enables improved Ki-67 prediction, supporting informed therapeutic decisions in HCC.