AI Models for Predicting TACE Response and Prognosis in Hepatocellular Carcinoma

Jiangqin Ma's research dataset contains models and results for predicting treatment response and prognosis after transarterial chemoembolization (TACE) in hepatocellular carcinoma (HCC) patients. The data includes performance metrics for clinical, radiomics, and deep learning models developed from gadoxetic acid-enhanced MRI scans of 160 patients from April 2018 to September 2024. An independent external validation cohort of 38 patients was also used.

Use Cases

• Benchmarking radiomics models for treatment response prediction based on 26 LASSO-selected features.
• Evaluating deep learning model performance for medical imaging tasks based on a ResNet50 architecture.
• Assessing the prognostic value of combined deep learning and radiomics scores for overall survival analysis.
• Comparing model calibration and clinical benefit using decision curve analysis (DCA) metrics.

Strengths

• Includes an external validation cohort of 38 patients, providing a robustness check.
• Reports specific model performance metrics, including AUC values for training (up to 0.96), test (up to 0.92), and validation (up to 0.86) sets.
• Dataset is openly licensed under CC-BY-4.0, facilitating reuse and sharing.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is very small (15.1 KB), suggesting it contains summary results or model parameters rather than raw patient data.

Provenance

Source

figshare

Collection Method

Retrospective study using gadoxetic acid-enhanced MRI scans from patients with Barcelona Clinic Liver Cancer (BCLC) stage A or B HCC.

Time Range

April 2018 to September 2024

Freshness

Last updated 2026-04-28 04:28:56; freshness should be verified.

Geography

null