基于人工智能的荧光法与传统流式法精子DFI检测的对比研究  

Artificial intelligence-based fluorescence method versus traditional flow cytometry in detection of sperm DNA fragmentation index

作　　者：开钧 汪珣 周雪[1] 朱来晴 李敏环 孙国海 韩友峰[1] 石亮[1] KAI Jun;WANG Xun;ZHOU Xue;ZHU Lai-qing;LI Min-huan;SUN Guo-hai;HAN You-feng;SHI Liang(Department of Andrology,Drum Tower Hospital Affiliated to Nanjing University School of Medicine,Nanjing,Jiangsu 210008,China)

机构地区：[1]南京大学医学院附属鼓楼医院男科,江苏南京210008

出　　处：《中华男科学杂志》2022年第3期227-231,共5页National Journal of Andrology

摘　　要：目的:探究基于人工智能(AI)识读技术的荧光法精子DNA碎片指数(DFI)检测在检测结果上与传统流式细胞法精子DFI检测的差异性以及AI荧光法DFI在检测方法上的可靠性。方法:338例门诊患者,对每例患者的精液样本分别采用流式法和AI荧光法进行DFI检测,分析两种方法检测结果的相关性及检测阳性率是否存在统计学显著差异;另外对每例患者的精液样本分别进行两次独立重复的AI荧光法DFI检测实验,观察AI荧光法DFI检测的稳定性。结果:流式法DFI与AI荧光法DFI两种检测的结果相关性较好(R;=0.7131),阳性率无统计学显著差异(P=0.10);对于极端精液样本(包括低浓度精液样本、黏稠精液样本),流式法DFI与AI荧光法DFI检测结果的相关性较差(R;=0.206 5)。AI荧光法DFI检测具有良好的技术稳定性(R;=0.967 1)。结论:AI荧光法DFI检测技术稳定性良好,检测结果与传统流式法DFI无显著差异,能够满足临床工作对于DFI检测的要求,具有较好的临床应用前景。Objective: To compare the result of the artificial intelligence(AI) recognition-based fluorescence method and that of traditional flow cytometry in the examination of the sperm DNA fragmentation index(DFI) and assess the reliability of the AI-based fluorescence detection. Methods: Using flow cytometry and the AI-based fluorescence method, we examined the sperm DFI in the semen samples collected from 338 outpatients. We analyzed the correlation between the results and compared the positive rates detected by the two methods. We repeated the AI-based fluorescence method twice for each semen sample to observe its technical stability in the detection of sperm DFI. Results: The result of flow cytometry was well correlated with that of the AI-based fluorescence method in the detection of sperm DFI(R;= 0.7131), but poorly correlated for low-concentration, sticky semen and some other extreme samples(R;= 0.2065). No statistically significant difference was found between the two methods in the positive rate of detection. The AI-based fluorescence method exhibited an excellent technical stability(R;= 0.9671). Conclusion: The AI-based fluorescence method has an excellent technical stability in the detection of sperm DFI and the result is not significantly different from that of traditional flow cytometry.

关 键 词：精子DNA碎片指数 染色质结构 流式细胞仪 人工智能技术 

分 类 号：R321.1[医药卫生—人体解剖和组织胚胎学]