Scholarlink

<p>Accurate estimation of body composition remains a critical task in personalized healthcare and clinical decision-making. However, existing measurement methods, such as Dual-Energy X-ray Absorptiometry (DXA) and Computed Tomography (CT), are often difficult to access and pose potential risks associated with ionizing exposure. In this study, we present a streamlined pipeline that replaces handcrafted feature engineering with end-to-end learning from voxel maps and demographic features. Our extended experiments demonstrate promising RMSE performance on real-world 3D scan datasets across multiple regional and total body composition values, validating the effectiveness of this voxel-demographic approach over traditional part-based feature descriptors. The proposed methodology represents a paradigm shift from manual feature extraction to automated pattern learning, demonstrating significant improvements in both accuracy and scalability for clinical body composition analysis.</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

Voxel-based Deep Regression for Enhanced Body Composition Estimation from 3D Body Scans

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月21日

查看全文
  • Boyuan Feng,
  • Ruting Cheng,
  • Yijiang Zheng,
  • Shuya Feng,
  • Ningshuo Bai,
  • Khashayar Vaziri,
  • James Hahn

摘要

Accurate estimation of body composition remains a critical task in personalized healthcare and clinical decision-making. However, existing measurement methods, such as Dual-Energy X-ray Absorptiometry (DXA) and Computed Tomography (CT), are often difficult to access and pose potential risks associated with ionizing exposure. In this study, we present a streamlined pipeline that replaces handcrafted feature engineering with end-to-end learning from voxel maps and demographic features. Our extended experiments demonstrate promising RMSE performance on real-world 3D scan datasets across multiple regional and total body composition values, validating the effectiveness of this voxel-demographic approach over traditional part-based feature descriptors. The proposed methodology represents a paradigm shift from manual feature extraction to automated pattern learning, demonstrating significant improvements in both accuracy and scalability for clinical body composition analysis.