Recent innovations in signal and image processing and data analysis in Raman spectroscopy

Ryabchykov, Oleg, Leibniz-Institut für Photonische Technologien e.V.; Cialla-May, Dana, Leibniz-Institut für Photonische Technologien e.V.; Silge, Anja, Friedrich-Schiller-Univ. Jena; Mostafapour, Sara; Mokari, Azadeh; Luo, Ruihao; Dehbozorgi, Pegah; Contreras, Jhonatan; Popp, Jürgen, Leibniz-Institut für Photonische Technologien e.V.; Bocklitz, Thomas, Leibniz-Institut für Photonische Technologien e.V.

doi:10.1117/3.100303.ch12

Book: Biophotonics and Biosensing: From Fundamental Research to Clinical Trials Through Advances of Signal and Image Processing

Editor(s): Andrea Armani; Tatevik Chalyan; David D. Sampson

Chapter Author(s): Oleg Ryabchykov, Dana Cialla-May, Anja Silge, Sara Mostafapour, Azadeh Mokari, Ruihao Luo, Pegah Dehbozorgi, Jhonatan Contreras, Jürgen Popp, Thomas Bocklitz

Published: 2024

https://doi.org/10.1117/3.100303.ch12

Author Affiliations +

Abstract

This chapter provides an overview of the data modeling techniques used for Raman spectroscopic data, including Raman spectra and Raman-based images, such as CARS images. Because of the different nature of both kinds of data, i.e., Raman spectra, which are 1D and images which are 2D, different methods for correction, standardizing, and analyzing are needed. The methods needed to analyze Raman spectra and Raman-based images fall into the category “signal processing” and “image processing.” The signal processing section described data pretreatment, including spike correction, and preprocessing, which included background correction and normalization, also current advances. Thereafter in the subsection “Image processing,” classical machine learning (ML) and deep learning (DL) methods for Raman spectroscopic data were summarized. It should be noted that the term classical machine learning refers to nondeep learning techniques, which is strongly overlapping with chemometrics. In the second main section of the chapter, image processing is introduced and advances for the application of Raman-based images are given. Image enhancement/standardization; classical machine learning (CML) for Raman-based imaging techniques; and deep learning (DL) for Raman-based imaging techniques are the subheadings that are discussed in this section, and new trends are highlighted.