五分类血液分析仪的智能光学系统设计  

Design of Intelligent Optical System in Five Classification Hematology Analyzer

作　　者：万里霞[1] 龙伟[1] 张星原[1] 卢斌[1] 

机构地区：[1]南昌大学信息工程学院,南昌330031

出　　处：《计算机测量与控制》2015年第6期2223-2225,共3页Computer Measurement &Control

基　　金：国家自然科学基金资助项目(61261011)

摘　　要：目前国内血液分析仪的白细胞五分类大多以硬件方式实现,且存在硬件结构复杂,制作成本高和过度依赖某些精密部件等问题;为简化五分类仪器的系统结构,提出了一种用于白细胞五分类的智能光学系统,该系统以全光学技术作为白细胞检测方式,采用VC6.0作为软件开发平台,建立了RBF神经网络的白细胞五分类识别算法模型,整个细胞识别和分类过程完全由软件实现,从而降低硬件复杂程度,减小了外界干扰因素的影响;实验结果:样机对LYM、MON、NEU、EOS、BAS的测试相对偏差分别为1.43%、4.41%、3.92%、2.94%、11.1%,满足了国家标准中的性能要求,故仪器整体的分类结果比较理想;经验证,文章提出的智能光学系统具有性能稳定可靠、抗干扰能力强的特点。At present, the WBC five classification in hematology analyzers are mostly implemented by hardware mode, and have the problems of complex hardware structure, high manufacture cost and excessive depend on some sophisticated components. Therefore, we put forward a intelligent optical system based on full--optical intelligent technology, the system take the full--optical technology as the WBC detection method, using VC6.0 as the software development platform, finally established the RBF neural network model for WBC recognition and five classification. The whole process of cells＇ recognition and classification are mainly realized by software, which simplify the hardware structure and reduce the influence of external factors. Results： the prototype test relative deviation is 1.43%, 4. 41%, 3.92%, 2.94%, 11.1 % which meet the relevant national standards for hematology analyzer, so the classification accuracy of the system is high. Conclusions： the intelligent optical system proposed has the features of consistently reliable performance and strong anti--interference ability.

关 键 词：血液分析仪 RBF神经网络 全光学技术 算法模型 识别准确度 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]