The identification of blood species using the correlation coefficient of sub-spectra based on Raman spectroscopy

For spectroscopic analysis, multivariable analysis methods such as partial least squares and support vector machine are usually required to extract the information of the interest such as blood species. However, for these algorithms, a number of spectra should be first selected as the training dataset to build the model and the cross-validation should be performed which is time-consuming and fussy. In this paper, the correlation coefficient of the sub-spectra based method was proposed to identify the blood species. Totally, 24 Raman spectra of blood samples originating from human volunteers and 50 Raman spectra of blood samples originating from rat, dog, rabbit, goose and duck were collected to demonstrate the feasibility of this new approach. The wavelet transform was used to decompose the Raman spectra of the blood with the Daubechies wavelet basis function (‘db3′) and the correlation coefficient for each sub-spectra was calculated. The experimental results for total 74 blood samples indicated that the third sub-spectra achieved by wavelet transform can be used to discriminate the blood species and the accuracy can be as high as 97.2% which can be accepted in the practical application.