基于过程分析技术细胞培养中多指标实时检测

Real-time monitoring of multiple properties in cell culture processes using processing analysis technology

为了实现悬浮动物细胞培养过程细胞生长状态和生化指标的实时在线检测,以中国仓鼠卵巢细胞(CHO)培养过程为研究对象,采用原位显微镜和在线拉曼光谱仪实时在线获得CHO细胞培养体系的图像和拉曼光谱图,并通过实时在线图像处理和基于遗传算法结合偏最小二乘法的化学计量学建模,分别实时获得培养体系中细胞的密度、直径分布、培养液中葡萄糖和乳酸浓度。图像处理结果与离线生化分析仪结果相一致;葡萄糖质量浓度校正模型的预测均方根误差(RMSEP)为0.085 g×L^(-1),预测决定系数RP2为0.999,乳酸质量浓度校正模型RMSEP为0.055 g×L^(-1),RP2为0.966。结果表明,上述检测方法可用于细胞培养过程的细胞密度、直径分布和2个生化指标的实时准确监测,减少了离线检测的繁琐和染菌风险。

In order to real-time and on-line monitor cell growth status and biochemical index of suspended animal cells,Chinese hamster ovary(CHO)cells were studied using in-situ microscopy and on-line Raman spectrometry.Based on image analysis and chemometric models obtained from genetic algorithm combined with partial least squares,cell density,size distribution,and glucose and lactic acid concentrations of culture media were acquired in real-time.The root mean square error(RMSEP)of glucose concentration model was 0.085 g×L^(-1) and the correlation coefficient RP2 was 0.999.The RMSEP and RP2 of lactic acid model were 0.055 g×L^(-1) and 0.966,respectively.These results show that the above method can be used to real-time and on-line monitor cell density,cell size distribution,and metabolic concentration.Compared with off-line monitoring methods,this method can avoid the risk of bacterial infection and complicated operation.