Scholarlink

<p>Uterine fibroids are a common indication for image-guided, uterus-preserving therapies such as focused ultrasound ablation surgery (FUAS), yet substantial interpatient variability in treatment difficulty limits reproducible analysis. Here, we present a curated multimodal dataset comprising structured clinical variables, continuous multisequence magnetic resonance imaging (MRI), procedural parameters, and quantitatively defined treatment outcome indicators from patients undergoing FUAS for uterine fibroids. The dataset integrates heterogeneous data modalities within a standardized and anonymized framework to support reproducible research and downstream reuse. To facilitate technical validation, we further provide a reference multimodal prediction pipeline that integrates dynamic MRI sequences with structured clinical data to estimate treatment difficulty and derive a rule-based FUAS recommendation label. Using this baseline implementation, the dataset supports joint regression and classification tasks, achieving a mean absolute percentage error of 0.265 and an area under the receiver operating characteristic curve of 0.931 on an independent test set. This openly available dataset provides a valuable resource for benchmarking and for multimodal data fusion research related to FUAS treatment of uterine fibroids.</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

A multimodal dataset and predictive model for the treatment of uterine fibroids with focused ultrasound ablation surgery

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月15日

查看全文
  • Qianru Zeng,
  • Ling Fan,
  • Shuaibin Liu,
  • Qian Wang,
  • Yu Zhao,
  • Min Zhou,
  • Li Ling

摘要

Uterine fibroids are a common indication for image-guided, uterus-preserving therapies such as focused ultrasound ablation surgery (FUAS), yet substantial interpatient variability in treatment difficulty limits reproducible analysis. Here, we present a curated multimodal dataset comprising structured clinical variables, continuous multisequence magnetic resonance imaging (MRI), procedural parameters, and quantitatively defined treatment outcome indicators from patients undergoing FUAS for uterine fibroids. The dataset integrates heterogeneous data modalities within a standardized and anonymized framework to support reproducible research and downstream reuse. To facilitate technical validation, we further provide a reference multimodal prediction pipeline that integrates dynamic MRI sequences with structured clinical data to estimate treatment difficulty and derive a rule-based FUAS recommendation label. Using this baseline implementation, the dataset supports joint regression and classification tasks, achieving a mean absolute percentage error of 0.265 and an area under the receiver operating characteristic curve of 0.931 on an independent test set. This openly available dataset provides a valuable resource for benchmarking and for multimodal data fusion research related to FUAS treatment of uterine fibroids.