Scholarlink

<p>Accurate prediction of early post-operative recurrence in hepatocellular carcinoma (HCC) is crucial for personalized treatment. This study introduces a multimodal framework integrating clinical parameters, pathological features, and CT radiomics, employing Gaussian Mixture Model (GMM) clustering for unsupervised radiomics filtering. Analyzing 205 HCC patients who underwent curative resection, we identified two distinct clusters and 50 discriminative features. The Clinical+Pathological+Radiomics feature set achieved superior performance (AUC: 0.909, 95% CI: 0.900−0.918), significantly outperforming other combinations. Our framework demonstrates potential for improving HCC recurrence prediction through comprehensive data integration and rigorous validation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multimodal feature fusion with GMM-filtered radiomics for predicting early HCC recurrence

  • 期刊论文

  • 发表日期：

    2026年5月19日

查看全文
  • Yuting Li,
  • Zhen Wang,
  • Zhenhu He,
  • Chenhong Guo,
  • Jiahao Liu,
  • Zhenwei Peng,
  • Ruhan Liu,
  • Pengfei Rong

摘要

Accurate prediction of early post-operative recurrence in hepatocellular carcinoma (HCC) is crucial for personalized treatment. This study introduces a multimodal framework integrating clinical parameters, pathological features, and CT radiomics, employing Gaussian Mixture Model (GMM) clustering for unsupervised radiomics filtering. Analyzing 205 HCC patients who underwent curative resection, we identified two distinct clusters and 50 discriminative features. The Clinical+Pathological+Radiomics feature set achieved superior performance (AUC: 0.909, 95% CI: 0.900−0.918), significantly outperforming other combinations. Our framework demonstrates potential for improving HCC recurrence prediction through comprehensive data integration and rigorous validation.