DeepRaman: Implementing surface-enhanced Raman scattering together with cutting-edge machine learning for the differentiation and classification of bacterial endotoxins.

Unlabelled: To classify raw SERS Raman spectra from biological materials, we propose , a new architecture inspired by the Progressive Fourier Transform and integrated with the scalogram transformation approach. Unlike standard machine learning approaches such as PCA, LDA, SVM, RF, GBM etc, functions independently, requiring no human interaction, and can be used to much smaller datasets than traditional CNNs. Performance of DeepRaman on 14 endotoxins bacteria and on a public data achieved an extraordinary accuracy of 99 percent.