SiameseNet based on multiple instance learning for accurate identification of the histological grade of ICC tumors.

Background: After hepatocellular carcinoma (HCC), intrahepatic cholangiocarcinoma (ICC) is the second most common primary liver cancer. Timely and accurate identification of ICC histological grade is critical for guiding clinical diagnosis and treatment planning.

Method: We proposed a dual-branch deep neural network (SiameseNet) based on multiple-instance learning and cross-attention mechanisms to address tumor heterogeneity in ICC histological grade prediction.

Radiomics-based biomarker for PD-1 status and prognosis analysis in patients with HCC.

Purpose: To investigate the impact of preoperative contrast-enhanced CT-based radiomics model on PD-1 prediction in hepatocellular carcinoma (HCC) patients.

Methods: The study included 105 HCC patients (training cohort: 72; validation cohort: 33) who underwent preoperative contrast-enhanced CT and received systemic sorafenib treatment after surgery. Radiomics score was built for each patient and was integrated with independent clinic radiologic predictors into the radiomics model using multivariable logistic regression analysis.