2015, Alebrahim, M.A., Krafft, C., Popp, J., El-Khateeb, Mohammad Y.

The structure and chemical features of the human dentin enamel junction (DEJ) were characterized using Raman spectroscopic imaging. Slices were prepared from 10 German, and 10 Turkish teeth. Raman images were collected at 785 nm excitation and the average Raman spectra were calculated for analysis. Univariate and multivariate spectral analysis were applied for investigation. Raman images were obtained based on the intensity ratios of CH at 1450 cm-1 (matrix) to phosphate at 960 cm-1 (mineral), and carbonate to phosphate (1070/960) ratios. Different algorithms (HCA, K-means cluster and VCA) also used to study the DEJ. The obtained results showed that the width of DEJ is about 5 pm related to univariate method while it varies from 6 to 12 μm based on multivariate spectral technique. Both spectral analyses showed increasing in carbonate content inside the DEJ compared to the dentin, and the amide I (collagen) peak in dentin spectra is higher than DEJ spectra peak.

2020, Korinth, F., Mondol, A.S., Stiebing, C., Schie, I.W., Krafft, C., Popp, J.

Shifted excitation Raman difference spectroscopy (SERDS) is a background correction method for Raman spectroscopy. Here, the difference spectra were directly used as input for SERDS-based classification after an optimization procedure to correct for photobleaching of the autofluorescence. Further processing included a principal component analysis to compensate for the reduced signal to noise ratio of the difference spectra and subsequent classification by linear discriminant analysis. As a case study 6,028 Raman spectra of single pollen originating from plants of eight different genera and four different growth habits were automatically recorded at excitation wavelengths 784 and 786 nm using a high-throughput screening Raman system. Different pollen were distinguished according to their growth habit, i.e. tree versus non-tree with an accuracy of 95.9%. Furthermore, all pollen were separated according to their genus, providing also insight into similarities based on their families. Classification results were compared using spectra reconstructed from the differences and raw spectra after state-of-art baseline correction as input. Similar sensitivities, specificities, accuracies and precisions were found for all spectra with moderately background. Advantages of SERDS are expected in scenarios where Raman spectra are affected by variations due to detector etaloning, ambient light, and high background.