都月, 孟晓辰, 祝连庆. 基于优化迭代算法的细胞荧光光谱解析[J]. 应用光学, 2019, 40(3): 461-467. DOI: 10.5768/JAO201940.0303001
引用本文: 都月, 孟晓辰, 祝连庆. 基于优化迭代算法的细胞荧光光谱解析[J]. 应用光学, 2019, 40(3): 461-467. DOI: 10.5768/JAO201940.0303001
DU Yue, MENG Xiaochen, ZHU Lianqing. Novel algorithm for analysis of cellular fluorescence spectrum based on optimization iteration[J]. Journal of Applied Optics, 2019, 40(3): 461-467. DOI: 10.5768/JAO201940.0303001
Citation: DU Yue, MENG Xiaochen, ZHU Lianqing. Novel algorithm for analysis of cellular fluorescence spectrum based on optimization iteration[J]. Journal of Applied Optics, 2019, 40(3): 461-467. DOI: 10.5768/JAO201940.0303001

基于优化迭代算法的细胞荧光光谱解析

Novel algorithm for analysis of cellular fluorescence spectrum based on optimization iteration

  • 摘要: 使用光谱测量方法进行细胞多色荧光分析时, 发射光谱会产生部分光谱重叠, 为定性和定量分析造成了一定的困难。为此, 提出基于优化迭代算法的细胞荧光光谱解析算法, 建立重叠峰模型并确定单峰顶点; 根据每次构造峰面积的大小, 重新确定构造峰的构造方式, 最终得到模拟峰的顶点及面积信息。利用该算法对高斯函数叠加形成的重叠峰进行解析, 并与常规方法进行对比, 结果表明优化迭代算法解析误差稳定在0.15%以内; 加入随机噪声后, 解析误差可稳定在0.85%以内, 均优于另外两种算法。此外, 计算了该算法下的迭代效率, 结果表明该算法较常规方法提高了32.2%。

     

    Abstract: When using the spectral measurement method for multi-color fluorescence analysis of cells, the emission spectrum can produce partial spectral overlap, which poses certain difficulties for qualitative and quantitative analysis. Aiming at the problem, a novel algorithm was developed to effectively decompose the overlapped spectral peaks based on optimization iteration. First, the separation model of overlapped peak was built up to obtain the peak vertex of overlapping peaks. And then, according to the size of each constructed peak area, the structure of the constructed peak was re-determined, and finally the vertex and area information of the simulated peak was obtained.Moreover, the algorithm was used to analyze the overlapping peaks formed by the superposition of Gaussian functions, and compared with the conventional methods. The results show that the analytical error of the optimized iterative algorithm is stable within 0.15%, while it can be stabilized within 0.85% after adding random noise, which are all better that the other two algorithms. In addition, the iterative efficiency under the algorithm was calculated, and the results show that the algorithm is 32.2% higher than the conventional method.

     

/

返回文章
返回