摘要
目的:提高血清质量检测水平,有效筛选问题血清,提高临床实验室检验结果的准确度。方法:使用工业相机对血清样本进行拍照,并利用图像识别的方法,获取血清图片,建立血清Yxy颜色特征图库,并使用SVM分类模型实现血清的识别。结果:在15411张血清样本上的测试结果,血清、溶血、脂血、黄疸样本的识别结果与人眼判断结果的符合率分别为:血清识别正确率为99.84%,溶血识别正确率为89.33%,脂血识别正确率为86.21%,黄疸识别正确率为83.87%。结论:血清检测系统的符合率满足使用要求,在一定程度上可以代替人眼做出判断。
Objective:To improve the serum quality testing level and detect the problem serum,in order to improve the accuracy of clinical laboratory test results.Methods:Using the image identification method to analyze the serum sample pictures from the industrial camera,in order to get the serum picture and develop the Yxy color feature library of serum.Using SVM classification model to recognize the serum.Results:Testing on 15411 serum specimens,the accordance rates of serum,hemolysis,lipidemia,icterus specimen between our methods and visual judgement were 99.84%,89.33%,86.21%,83.87%.Conclusion:The accordance rates of serum detection system can accomplish the request for use.And the serum detection system can replace visual judgement to screen the trouble serum specimens to some degree.
作者
侯剑平
王超
赵万里
段忆芮
HOU Jian-ping;WANG Chao;ZHAO Wan-li;DUAN Yi-rui(Autobio Labtec Instruments Co.,Ltd.,Henan Zhengzhou 450016)
出处
《中国医疗器械信息》
2020年第17期30-33,共4页
China Medical Device Information
