RT期刊文章SR电子T1面部缺陷的自动检测使用机器学习(S40.001)摩根富林明神经病学神经学乔FD Lippincott Williams &威尔金斯SP S40.001 VO 90 15 A1补充奥马尔·乌里韦A1马首页克麦当劳A1燕壮族A1虹膜林A1威廉丹尼尔Arteaga A1 Dalrymple A1布拉德福德worral A1 Gustavo罗德A1安德鲁Southerland年2018 UL //www.ez-admanager.com/content/90/15_Supplement/S40.001.abstract AB目的:本试验研究的目的是确定一种机器学习算法可以准确地检测面部疲软的静态图像。背景:早期识别和治疗中风的改善结果。院前筛查工具为急性中风的早期发现提供承诺但有不一致的性能。自动化检测工具可以减少inter-operator可变性和操作员错误。我们假设机器学习算法可以协助检测病理面部弱点使用计算机视觉分析。设计/方法:两位高级神经病学居民得分(n = 333)首页图像显示正常的微笑和面部的弱点在5范围内。只有图像评价认识提高,评级机构可能正常或可能包括异常进行分析。提取面部地标定位算法使用一个开源的脸检测和地标。惩罚线性判别分析方法被用来处理数据的再分类器算法。我们使用5倍交叉验证方案和计算精度,灵敏度和特异性评估性能。结果:199年的图像分析,18图片被排除由于局限在面部具有里程碑意义的提取。剩余的181张照片,87年被评为可能正常,94年被评为可能有面部的弱点。算法执行的准确率达到了93.8% (95% CI[93.4 - -94.2]), 95.8%的敏感性(95.4 - -96.2)和93.8%的特异性(93.4 - -94.2)。Conclusions: In this pilot study, we demonstrate that machine learning algorithms can accurately identify facial weakness in static images. These results support further evaluation of machine learning algorithms in the early detection of acute stroke symptoms. Future research will examine video analysis algorithms in detecting facial weakness and other signs in acute stroke patients.Disclosure: Dr. Uribe has nothing to disclose. Dr. McDonald has nothing to disclose. Dr. Zhuang has nothing to disclose. Dr. Lin has nothing to disclose. Dr. Arteaga has nothing to disclose. Dr. Dalrymple has nothing to disclose. Dr. Worrall has nothing to disclose. Dr. Rohde has nothing to disclose. Dr. Southerland has nothing to disclose.