摘要
目的:构建面部损容性色素性疾病的人工智能诊断模型,实现面部损容性色素性疾病的人工智能辅助诊断。方法:利用雀斑、黄褐斑、颧部褐青色痣、日光性黑子、太田痣、咖啡斑、白癜风的单反相机图像和YOLO(You Only Look Once)v5算法建立诊断模型,比较模型与我院以及基层医院皮肤科医生的诊断结果,评价模型的性能。结果:对于350张单反相机图像的诊断结果显示,模型对七种面部损容性色素性疾病的平均诊断准确率为94.29%,高于基层医院皮肤科医生(81.43%),与我院皮肤科医生(97.48%)相当。结论:面部损容性色素性疾病人工智能诊断模型显示出较好的性能,提供了一种较为客观、便捷的面部损容性色素性疾病的辅助诊断方法。
Objective: To construct an artificial intelligence diagnostic model for facial disfiguring pigmentation diseases, and to implement the artificial intelligence-assisted diagnosis of facial disfiguring pigmentation diseases. Methods: The SLR camera images of freckles, melasma, acquired nevi of Ota, solar lentigo, nevus of Ota, café-au-lait spots, and vitiligo, as well as YOLO(You Only Look Once) v5 algorithm, was used to establish a diagnostic model. The model was then compared with the diagnosis made by the dermatologists from our hospital and other community hospitals to evaluate the performance of the model. Results: The diagnostic results of 350 SLR camera images show that the average diagnostic accuracy rate of the model for 7 types of facial disfiguring pigmentation diseases is 94.29%, which is higher than that of the dermatologists from community hospitals(81.43%), and is equivalent to that of the dermatologist from our hospital(97.48%). Conclusion: The artificial intelligence diagnostic model of facial disfiguring pigmentation diseases shows a good performance, it provides a relatively objective and convenient diagnosis method for facial disfiguring pigmentation diseases.
作者
郭丽芳
丁徽
杨寅
葛一平
林彤
GUO Li-fang;DING Hui;YANG Yin;GE Yi-ping;LIN Tong(Department of Laser Surgery,Institute of Dermatology,CAMS and PUMC,Nanjing 210042,China)
出处
《临床皮肤科杂志》
CAS
CSCD
北大核心
2021年第9期536-539,共4页
Journal of Clinical Dermatology
基金
中国医学科学院医学与健康科技创新工程项目(CIFMS-2017-I2M-1-017、CIFMS-2018-I2-AI-018)。
关键词
色素性疾病
损容性
人工智能
诊断模型
pigment facial skin lesions
artificial intelligence
diagnostic model
