Transmission versus reflection spectroscopy for discrimination of human and nonhuman blood

Abstract

Discrimination of human blood is significantly important for import-export ports or inspection and quarantine departments. Blood products is very precious. Open sampling may pollute blood samples. Visible and near infrared spectroscopy method has been proved to be effective for human blood discrimination. This research compared the performance of reflection spectroscopy and transmission spectroscopy on discrimination the origin of whole blood samples. Partial Least Squares-Discriminant Analysis (PLS-DA) models were built with both transmission and reflection spectra. Experimental results showed that the two methods both worked well in the prediction of blind test (blood species within the calibration set). But for the prediction of external test (blood species outside the calibration set), the performance of the model based on reflection spectroscopy was better than that based on transmission spectroscopy. Therefore, the calibration model based on reflection spectroscopy has a better extrapolation ability. The most possible reason may be that the location repeatability requirement of anticoagulation for reflection spectroscopy was lower than that for transmission spectroscopy. Therefore, reflection spectroscopy was recommended in the further human blood discrimination researches.