基于卷积神经网络的X线片下肢关节角度识别算法
Lower limb joint angle calculation algorithm based on convolutional neural network in X-ray films
刘静妮 1盛玉武 1赵长秀 1牛存良 2黄国源 2许长栋 1赵姗姗 1陈彬1
作者信息
- 1. 武威市人民医院放射科,甘肃武威 733000
- 2. 武威市人民医院骨科,甘肃武威 733000
- 折叠
摘要
提出一种基于卷积神经网络的X线片下肢关节角度识别算法,首先在X线片中使用Yolov5目标检测模型来识别特定类别的感兴趣区域,并使用U-Net模型进行热图回归来识别关键特征点,最后进行下肢关节角度的计算.研究结果表明,本文提出的算法相比于之前的算法精度更高,结果准确可靠,为临床研究和实践提供参考.
Abstract
A convolutional neural network-based algorithm is proposed for calculating lower limb joint angle in X-ray films.After identifying the region of interest of a specific category in X-ray films through Yolov5 object detection model,U-Net model is used to perform heat map regression for identifying the key feature points,and then the lower limb joint angle is calculated.The results show that the proposed algorithm has higher accuracy than the previous algorithms and can obtain accurate and reliable results,providing references for clinical research and practice.
关键词
卷积神经网络/目标检测/特征点定位/下肢力线Key words
convolutional neural network/object detection/feature point localization/lower limb power line引用本文复制引用
基金项目
甘肃省武威市科技计划(WW23B02SF056)
出版年
2024
NETL
NSTL
万方数据