Radiomics Features for Differentiating Liver Tumors from Diffusion MRI

A dataset of 1,197 radiomics features extracted from diffusion MRI maps for 232 patients with liver tumors. The data was used to construct logistic regression models for differentiating hepatocellular carcinoma from hepatic hemangioma. The dataset was created by Jingwen Yu and last updated in May 2026.

Use Cases

• Training a binary classifier to differentiate hepatocellular carcinoma from hepatic hemangioma based on radiomics features.
• Comparing the diagnostic performance of models built from different MRI-derived maps (b0, b50, ADC, DDVD).
• Feature selection analysis to identify the most informative radiomics signatures for liver tumor characterization.

Strengths

• Dataset includes 232 patient cases (104 HCC, 128 HG), providing a substantial sample for model development.
• Model performance is validated with reported AUC values (e.g., 0.977 in the test cohort) and statistical comparisons.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect the specific patient cohort and imaging protocols used in the study.

Provenance

Source

figshare

Collection Method

Retrospective study enrolling patients with pathologically confirmed HCC and clinically diagnosed HG.

Freshness

Last updated 2026-05-28 04:43:21; freshness should be verified.