良性阵发性位置性眩晕的辅助诊断-《中国医学物理学杂志》

文章信息/Info

作者:: 刘雪冰1; 2; 高永彬1; 2; 1. 上海工程技术大学电子电气工程学院，上海 201600； 2.上海工程技术大学智慧医疗研究所，上海 201600

Author(s):: LIU Xuebing1; 2; GAO Yongbin1; 2; 1. School of Electronic and Electrical Engineering, Shanghai University of Engineering Science, Shanghai 201600, China 2. Smart Medical Research Institute, Shanghai University of Engineering Science, Shanghai 201600, China

Keywords:: Keywords: benign paroxysmal positional vertigo binaryzation thresholding template matching

摘要:: 目的：设计一种利用视觉虹膜识别概念和神经网络来诊断良性阵发性位置性眩晕（BPPV）的方法，从而辅助医生进行诊断。方法：通过计算机视觉二值化和阈值化技术从BPPV患者眼球运动视频中定位瞳孔和虹膜，捕获眼球运动轨迹。在眼球运动过程中使用虹膜模板匹配方法提取瞳孔和眼球运动的特征信息。将获得的眼动数据输入设计的神经网络进行BPPV诊断。结果：通过计算机视觉技术的实时检测和网络学习特征的能力，能有效捕捉并描述眼球运转轨迹，输出当前眼动轨迹信息，辅助医生做出更准确、更迅速的决策。结论：使用这种相对简单、无创且精确的方法来测量扭转眼球运动解决了长期存在的阻碍视觉和眼球运动的问题，能广泛应用于基础和临床研究或诊断测试。

Abstract:: Abstract: Objective To diagnose benign paroxysmal positional vertigo (BPPV) using visual iris recognition concept and neural network for assisting doctors in diagnosis. Methods The pupil and iris were located from the eye movement video of BPPV patients using computer vision binarization and thresholding techniques to capture the eye movement trajectory. During eye movement, iris template matching method was used to extract the feature information of the pupil and eye movement. The obtained data was input into the designed neural network for BPPV diagnosis. Results Through the real-time detection of computer vision technology and the feature learning of network, the proposed method effectively captured and described the eye movement trajectory, and output the eye movement trajectory, thereby assisting doctors in more accurate and faster decision-making. Conclusion Using the proposed method which is relatively simple, non-invasive and accurate to measure torsional eye movements solves the long-standing problems that hinder vision and eye movement, and it can be widely used in basic and clinical researches or diagnostic testing.