Scholarlink

<p>The objectives of this study are to identify tumor subregions and develop a habitat-driven, MRI-based predictive radiomics model for risk stratification and further assess the spatial heterogeneity within prostate cancer (PCa). This retrospective study included 344 patients with histologically confirmed PCa who underwent preoperative MRI. Patients were categorized into low-to-intermediate and high-risk groups according to the EAU-EANM-ESTRO-ESUR-SIOG guidelines. Intensity clustering of T2-weighted imaging (T2WI) and dynamic contrast-enhanced MR images were applied to partition the whole gland and tumor into multiple habitat subregions. Predictive models based on subregional and conventional radiomic features were established using logistic regression, and their performances were assessed through receiver operating characteristic curve analysis. Further analysis explored the correlation between the features of tumor subregions and risk-related clinical-pathological parameters, as well as the spatial heterogeneity in tumor subregions. The tumor subregion model outperformed the whole-gland model. Specifically, the tumor subregion 3 model derived from the T2WI (HD_T2W_TR3) demonstrated superior efficacy in the validation set, achieving an AUC of 0.84 (95% confidence interval 0.766–0.923). Habitat-radiomic features were strongly correlated with prostate-specific antigen and T-stage. Our research results emphasize the potential benefits of a habitat-driven model for risk stratification prediction of PCa. These findings provide valuable information in identifying high-risk tumor subregions.</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

Precision Risk Stratification in Prostate Cancer: Unveiling MRI-Based Habitat-Driven Radiomics Model for Enhanced Treatment Guidance

  • 期刊论文

  • 发表日期：

    2025年12月9日

查看全文
  • Yang Qiu,
  • Huan Liu,
  • Xin Shu,
  • Yunfan Liu,
  • Xiaofeng Qiao,
  • Xiaojing He

摘要

The objectives of this study are to identify tumor subregions and develop a habitat-driven, MRI-based predictive radiomics model for risk stratification and further assess the spatial heterogeneity within prostate cancer (PCa). This retrospective study included 344 patients with histologically confirmed PCa who underwent preoperative MRI. Patients were categorized into low-to-intermediate and high-risk groups according to the EAU-EANM-ESTRO-ESUR-SIOG guidelines. Intensity clustering of T2-weighted imaging (T2WI) and dynamic contrast-enhanced MR images were applied to partition the whole gland and tumor into multiple habitat subregions. Predictive models based on subregional and conventional radiomic features were established using logistic regression, and their performances were assessed through receiver operating characteristic curve analysis. Further analysis explored the correlation between the features of tumor subregions and risk-related clinical-pathological parameters, as well as the spatial heterogeneity in tumor subregions. The tumor subregion model outperformed the whole-gland model. Specifically, the tumor subregion 3 model derived from the T2WI (HD_T2W_TR3) demonstrated superior efficacy in the validation set, achieving an AUC of 0.84 (95% confidence interval 0.766–0.923). Habitat-radiomic features were strongly correlated with prostate-specific antigen and T-stage. Our research results emphasize the potential benefits of a habitat-driven model for risk stratification prediction of PCa. These findings provide valuable information in identifying high-risk tumor subregions.