Figure 2

PCA was applied to the background corrected spectra. The scores plot of the first two PCs indicates the changes corresponding to the DOX treatment (a). The corresponding described variances by PC1 and PC2 are mentioned in the axis label (a). (b) Loadings of the PCA model. The scores corresponding to the first two loadings, PC1 and PC2, were employed for the SVM model. Afterwards, the SVM model was applied to the dataset. (c) The prediction of PCA-SVM separates viable and non-viable cells. The color-coding is done based on the PCA-SVM model prediction whereas the model regions are separated by the decision boundary – solid black line. (d) The mean spectra of corresponding predicted classes by the SVM model also show pronounced spectral differences in several band positions.