Scholarlink

<p>Congenital Radio-Ulnar Synostosis (CRUS) causes difficulty in forearm rotation, which is treated using osteotomy. Effective preoperative planning of osteotomy requires automatic and objective quantification of deformity angles. In this paper, an automatic Congenital Radio-Ulnar Synostosis deformity evaluation method (CRUS-DE) is proposed. Initially, a method using threshold-layer tracking (TLT) and the segment anything model (SAM) is designed to recognize and segment the forearm from CT images. Subsequently, the Gaussian Process Morphable Model, in conjunction with refinement based on anatomical characteristics is developed to accurately identify forearm landmarks. Finally, the model automatically estimates deformity angles for quantitative assessment of CRUS. The forearm landmarks and deformity angles were successfully obtained from CT images based on CRUS-DE with average errors ranging from 0.98 to 1.55&#xa0;mm and from 0.7° to 2.4°, respectively. No significant differences existed between the automatic method and manual method. The CRUS-DE, as an explainable method, presented excellent performance in the quantification of deformity angles. This method can be used in preoperative planning and postoperative evaluation of osteotomy for forearm deformities.</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

An automatic congenital radio-ulnar synostosis deformity evaluation method (CRUS-DE): integrating TLT-SAM and GPMM-R for landmark identification

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月28日

查看全文
  • Lu Liu,
  • Ying Cui,
  • Tianfeng Zhou,
  • Shanlin Chen,
  • Yubing Guo,
  • Xinhua Zhou

摘要

Congenital Radio-Ulnar Synostosis (CRUS) causes difficulty in forearm rotation, which is treated using osteotomy. Effective preoperative planning of osteotomy requires automatic and objective quantification of deformity angles. In this paper, an automatic Congenital Radio-Ulnar Synostosis deformity evaluation method (CRUS-DE) is proposed. Initially, a method using threshold-layer tracking (TLT) and the segment anything model (SAM) is designed to recognize and segment the forearm from CT images. Subsequently, the Gaussian Process Morphable Model, in conjunction with refinement based on anatomical characteristics is developed to accurately identify forearm landmarks. Finally, the model automatically estimates deformity angles for quantitative assessment of CRUS. The forearm landmarks and deformity angles were successfully obtained from CT images based on CRUS-DE with average errors ranging from 0.98 to 1.55 mm and from 0.7° to 2.4°, respectively. No significant differences existed between the automatic method and manual method. The CRUS-DE, as an explainable method, presented excellent performance in the quantification of deformity angles. This method can be used in preoperative planning and postoperative evaluation of osteotomy for forearm deformities.