Multiphase MRI radiomics model for predicting microvascular invasion in HCC: Development and clinical validation.

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Accurate preoperative prediction of microvascular invasion (MVI) is critical for hepatocellular carcinoma (HCC) treatment planning.
• Hepatocellular carcinoma (HCC) is a primary liver cancer with significant morbidity and mortality.
• Microvascular invasion (MVI) is a key prognostic factor in HCC, influencing treatment decisions and patient outcomes.

Purpose of the Study:

• To develop and validate a multi-phase magnetic resonance imaging (MRI)-based radiomics model for predicting MVI in HCC patients.
• To evaluate the performance of the radiomics model in comparison to single, two, and three-phase models.
• To establish a non-invasive tool for improving preoperative assessment of MVI in HCC.

Main Methods:

• Retrospective study of 110 HCC patients with preoperative multi-phase MRI (non-contrast, arterial, portal, hepatobiliary).
• Extraction of radiomics features from four MRI phases.
• Feature selection using least absolute shrinkage and selection operator (LASSO) regression and evaluation of five machine learning classifiers.

Main Results:

• The four-phase radiomics model with logistic regression achieved optimal performance (AUC = 0.896 training, 0.889 validation).
• The multi-phase model outperformed single, two, and three-phase models in the validation cohort (AUCs 0.789, 0.815, 0.848 respectively).
• In the validation cohort, the model demonstrated balanced performance with sensitivity, specificity, accuracy, and precision all at 0.857.

Conclusions:

• A multi-phase MRI-based radiomics model significantly improves MVI prediction accuracy in HCC patients.
• This non-invasive approach offers a valuable tool for enhancing preoperative assessment and guiding treatment planning in HCC.
• The developed model has the potential to optimize patient management and improve clinical outcomes for HCC.