Wang Bolu, Wu Xiaomei
Fudan University, Shanghai, 200433.
Zhongguo Yi Liao Qi Xie Za Zhi. 2022 Jan 30;46(1):57-62. doi: 10.3969/j.issn.1671-7104.2022.01.012.
This paper reviews some recent studies on the recognition and evaluation of facial paralysis based on artificial intelligence. The research methods can be divided into two categories: facial paralysis evaluation based on artificial selection of patients' facial image eigenvalues and facial paralysis evaluation based on neural network and patients' facial images. The analysis shows that the method of manual selection of eigenvalues is suitable for small sample size, but the classification effect of adjacent ratings of facial paralysis needs to be further optimized. The neural network method can distinguish the neighboring grades of facial paralysis relatively well, but it has a higher requirement for sample size. Both of the two methods have good prospects. The features that are more closely related to the evaluation scale are selected manually, and the common development direction may be to extract the time-domain features, so as to achieve the purpose of improving the evaluation accuracy of facial paralysis.
本文综述了一些近期基于人工智能对面部瘫痪进行识别和评估的研究。研究方法可分为两类：基于人工选择患者面部图像特征值的面瘫评估和基于神经网络与患者面部图像的面瘫评估。分析表明，人工选择特征值的方法适用于小样本量，但面瘫相邻等级的分类效果有待进一步优化。神经网络方法能较好地区分面瘫的相邻等级，但对样本量要求较高。这两种方法都有良好的前景。人工选择与评估量表相关性更强的特征，共同的发展方向可能是提取时域特征，以达到提高面瘫评估准确性的目的。
