Analysis Of Near-infrared Spectra Data Of Osteonecrosis Of Femoral Head

Objective:The purpose of this study is trying to detect the metabolite changes using Near-Infrared Spectroscopy while starting a certain chemometric classification model to discriminate osteonecrosis from normal tissues.Methods:A total of 170 surgical specimens were prepared,including 62 ONBT tissues,66 ANBT tissues and 42 NBT tissues.NIR spectra were recorded by an integrating sphere.The experiment data set was divided into three subsets,i.e.,the training set,validation set,and test set.Principal component analysis(PCA)was used for exploratory analysis firstly,Successive projection algorithm-linear discriminant analysis(SPA-LDA)was used to compress variables and build the discriminant model,partial least square-discriminant analysis(PLS-DA)was used as the reference..Results : NIR spectroscopy combined with chemometric classification model is feasible for this research while an expected goal has been reached.For all the 3classification models,the model(ONBT verus NBT 、 ONBT verus ANBT)have achieved better performance than model(ANBT verus NBT).Compared to PLS-DA,SPA-LDA provided a more parsimonious model with better sensitivity and speficity.While the model(ONBT verus ANBT)can classify NBT exactly,the model(ONBT verus NBT)can hardly classify ANBT tissues.Conclusions: It indicated that NIR spectroscopy combined with chemometric classification model was a feasible aid tool for discriminating ONFH from normal tissues with an ideal sensitivity and specificity.