Scholarlink

Background <p>To develop and validate a contrast-enhanced ultrasound (CEUS)-based habitat model for noninvasive prediction of epidermal growth factor receptor (EGFR) mutation status in patients with peripheral non-small cell lung cancer (NSCLC).</p> Methods <p>This retrospective study included 187 patients with NSCLC confirmed by histopathology from April 2021 to February 2025. All patients underwent CEUS of the lung before biopsy. Patients for whom complete EGFR gene testing results were available were randomly divided into a training set and a test set at a ratio of 8:2. Habitat imaging was used to differentiate the tumor into distinct regions, and then an unsupervised clustering method was used to extract and analyze habitat features to establish Habitat_Model. The Shapley additive explanations (SHAP) method was used to improve the interpretability of the model. In addition, Rad_Model based on radiomics features of tumors was constructed. Finally, a combined model was established by combining the habitat features and clinical–radiological indicators with logistic regression analysis. The predictive performance was evaluated using receiver operating characteristic curve (ROC), calibration, and decision curve analysis (DCA).</p> Results <p>The areas under the curve (AUCs) of Habitat_Model and Rad_Model were 0.898 and 0.857 in the training set, and 0.784 and 0.649 in the testing set, respectively. Habitat_Model showed excellent performance. Incorporating clinical–radiological indicators via the combined model slightly improved its performance, resulting in AUC values of 0.904 and 0.812 for the training and testing sets, respectively. The calibration curves and DCA exhibited excellent fit for the combined model, while providing great clinical net benefit.</p> Conclusions <p>Our habitat model demonstrates a good capacity for predicting EGFR mutation status in peripheral NSCLC, providing valuable noninvasive reference for clinical pathways on targeted therapy.</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

Habitat model based on CEUS for noninvasive prediction of EGFR mutation status in peripheral NSCLC

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月26日

查看全文
  • Jingtong Zeng,
  • Liyan Wei,
  • Hengfei Chen,
  • Yingzi Liang,
  • Fangyi Huang,
  • Tian Qin,
  • Yong Gao,
  • Xinhong Liao

摘要

Background

To develop and validate a contrast-enhanced ultrasound (CEUS)-based habitat model for noninvasive prediction of epidermal growth factor receptor (EGFR) mutation status in patients with peripheral non-small cell lung cancer (NSCLC).

Methods

This retrospective study included 187 patients with NSCLC confirmed by histopathology from April 2021 to February 2025. All patients underwent CEUS of the lung before biopsy. Patients for whom complete EGFR gene testing results were available were randomly divided into a training set and a test set at a ratio of 8:2. Habitat imaging was used to differentiate the tumor into distinct regions, and then an unsupervised clustering method was used to extract and analyze habitat features to establish Habitat_Model. The Shapley additive explanations (SHAP) method was used to improve the interpretability of the model. In addition, Rad_Model based on radiomics features of tumors was constructed. Finally, a combined model was established by combining the habitat features and clinical–radiological indicators with logistic regression analysis. The predictive performance was evaluated using receiver operating characteristic curve (ROC), calibration, and decision curve analysis (DCA).

Results

The areas under the curve (AUCs) of Habitat_Model and Rad_Model were 0.898 and 0.857 in the training set, and 0.784 and 0.649 in the testing set, respectively. Habitat_Model showed excellent performance. Incorporating clinical–radiological indicators via the combined model slightly improved its performance, resulting in AUC values of 0.904 and 0.812 for the training and testing sets, respectively. The calibration curves and DCA exhibited excellent fit for the combined model, while providing great clinical net benefit.

Conclusions

Our habitat model demonstrates a good capacity for predicting EGFR mutation status in peripheral NSCLC, providing valuable noninvasive reference for clinical pathways on targeted therapy.