基于化学计量法的FTIR鉴别心源性猝死

目的联合应用傅里叶变换红外光谱技术及化学计量法对心源性猝死的大鼠进行鉴别分析。方法注射药物或空气分别诱导大鼠的心源性猝死及空气栓塞死亡模型。应用Nicolet i Z10 FTIR光谱仪采集血清样本的红外光谱,联合OMNIC及Unscrambler软件,基于光谱不同分子吸收区间建立偏最小二乘回归模型。结果基于指纹区(1 200~900cm-1)建立PLS模型,均方根误差和决定系数分别是0.068 3、0.981 3,预测均方根偏差和预测决定系数分别是0.104 8、0.956 1;基于蛋白质酰胺区间(1 720~1 480cm-1)建立PLS模型,RMSE和R2分别是0.058 6,0.986 2,RMSEP和预测R~2分别是0.079 4、0.974 7。综合分析两种模型,蛋白质酰胺区间的RMSEP更小,且预测R2更大,提示其预测效果略好于指纹区间。结论联合FTIR及化学计量法,基于蛋白质及指纹区的分子特征均可有效鉴别大鼠的心源性猝死,其中两种不同死因引起的蛋白质差异更为显著。

Differential diagnosis of sudden cardiac death by FTIR based on stoichiometry

Objective To differentially diagnose the sudden cardiac death of rats combined by FTIR and Stoichiometry. Methods The sudden cardiac death and death caused by air embolism models of rats were developed by injection of medication and air, respectively. The spectra data of serum samples were collected by FTIR(Nicolet iZ10) spectroscopy. The spectra were preprocessed and used to build PLS regression models on different regions of molecular absorption by OMNIC and Unscrambler software. Results The RMSE and R2 of the PLS model based on the fingerprint region(1200~900cm-1) were 0.0683 and 0.9813, respectively, and the RMSEP and predicted R2 were 0.1048 and 0.9561, respectively. Meanwhile, the RMSEP and R2 of the PLS model based on amides region(1720~1480cm-1) were 0.0586 and 0.9862, respectively, and the RMSEP and predicted R2 were 0.0794 and 0.9747, respectively. Comprehensively analyzed the present models, the model based on amides region had lower RMSEP and higher predicted R2, which could mean better predictive effect. Conclusion It is useful to differentially diagnose the SCD based on the molecular characteristics in protein and fingerprint regions combined by FTIR and Stoichiometry, thereinto, the differences of protein caused by the two causes of death are more significant.