在克罗地亚人口中使用适应的PointNet++网络对3D舌骨模型进行基于深度学习的性别估计
Ivan Jerković1, Željana Bašić2, Ivana Kružić1
1Faculty of Forensic Sciences, University of Split, Split, Croatia.
Scientific reports
|July 2, 2025
概括
深度学习使用CT扫描的3D舌骨模型准确估计性别. 这种方法有助于法医人类学,通过识别骨遗骸中的基于性别的形态模式.
科学领域:
- 法医人类学 法医人类学
- 生物医学工程 生物医学工程
- 计算机科学 计算机科学
背景情况:
- 从骨遗骸中估计性别在法医人类学中至关重要.
- 状腺骨形态学为性别确定提供了潜在的可能性.
- 先进的计算方法可以分析复杂的3D骨架数据.
研究的目的:
- 开发和评估一个深度学习方法,使用3D状骨模型来估计性别.
- 评估深度学习模型的准确性和可解释性.
- 探索这种方法在法医人类学应用中的潜力.
主要方法:
- 利用来自于克罗地亚人口中202名个人 (101名男性,101名女性) 的计算机断层扫描 (CT) 获得的3D状骨模型.
- 将CT衍生网格转换为2048点云,用于与适应的PointNet++网络进行处理.
- 员工无监督集群和监督分类 (支持矢量机器) 用于性别估计.
主要成果:
- 无监督的聚类在识别基于性别的形态模式方面实现了87.10%的准确性.
- 监督分类在测试组上产生了88.71%的准确性 (MCC = 0.7746).
- 解释性分析显示出明显的形态差异:雄性具有更大的U形舌头;雌性具有更小,更开放的结构.
结论:
- 深度学习方法有效地捕捉了雄骨形态中的性别差异.
- 该方法具有数据效率和可解释性,显示了在有限的骨遗骸中估计性别的潜力.
- 这个计算工具为法医人类学和相关领域提供了实用的见解.
相关概念视频
Classification of Bones
7.4K
The bones of the human skeletal system are of varied shapes, sizes, and functions. They can be classified based on their shape and function into four major classes: long bones, short bones, flat bones, and irregular bones. Some classifications include a fifth type, the sesamoid bones, as a separate class, whereas others categorize them under short bones.
Long and Short Bones
The appendicular skeleton, particularly the upper and lower limbs, is primarily made of long and short bones. The...
Long and Short Bones
The appendicular skeleton, particularly the upper and lower limbs, is primarily made of long and short bones. The...
7.4K
The Hyoid Bone
3.0K
The hyoid bone is a small U-shaped bone located in the upper neck at the level of the inferior mandible, with its tips pointing posteriorly. It does not directly articulate with any other bone in the body. The hyoid acts as the attachment site for the tongue, the larynx, and the pharynx. It is held in position by a series of small muscles attached from above or below. These muscles help to move the hyoid up/down or forward/back in coordination with movements of the tongue, larynx, and pharynx...
3.0K


