The purpose of this study is to develop a universal algorithm for analyzing the spectrograms of Raman scattering (Raman-spectroscopy) using intelligent data processing methods. It is found that the optimal algorithm is the calculation of the baseline spectrum by iterative polynomial regression, subsequent frequency analysis of the spectrum, determination of the main components of the spectrum, and machine learning. The reliability of identification and classification of normal tissue and tissue of a malignant tumor was 97.5 to 98 %.
deep learning, Raman scattering, Raman spectroscopy, malignant tumor, frequency analysis
1. Li Q.-B., Wang W., Liu Ch.-H., Zhanga G.-J. Discrimination of breast cancer from normal tissue with Raman spectroscopy and chemometrics. J. Applied Spectroscopy, 2015, vol. 82, pp. 450-455.
2. Jermyn M., Desroches J., Aubertin K. [et al.] A review of Raman spectroscopy advances with an emphasis on clinical translation challenges in oncology. Phys. Med. Biol., 2016, vol. 61, no. 23, pp. R370-R400.
3. ur Rehman I., Movasaghi Z., Rehman Sh. Vibrational spectroscopy for tissue analysis. CRC Press, 2012, 356 p.
4. Zhu J., Zhou J., Guo J. [et al.] Surface-enhanced Raman spectroscopy investigation on human breast cancer cells. Chemistry Central Journal., 2013, pp. 7-37.
5. Harmsen S., Wall M.A., Huang R., Kircher M.F. Cancer imaging using surface-enhanced resonance Raman scattering nanoparticles. Nat. Protoc., 2017, vol. 12, pp. 1400-14.
6. Harris A.T., Lungari A., Needham C.J. [et al.] Potential for Raman spectroscopy to provide cancer screening using a peripheral blood sample. Head Neck Oncol., 2009, vol. 1, no. 1, art. 34.
