Raman hyperspectral imaging and the implementation of machine learning algorithms: an innovative approach for molecular mapping of materials

Nonlinear Raman scattering spectroscopy is an optical method used to detect vibrational modes of target molecules, enabling the analysis and detection of chemical species at the molecular scale. When combined with a confocal optical microscope, this system can scan the sample surface point-by-point and record the molecular information present. By processing the raw spectral data and spatial coordinates, two-dimensional hyperspectral images are generated, containing nonlinear Raman scattering information for each pixel. Advanced statistical methods in artificial intelligence and machine learning are then used to extract information about the spatial distribution of molecules and chemical phases with high sensitivity and accuracy. In this study, an integrated optical setup consisting of a confocal optical microscope and a Raman spectroscopic system was used to scan the surface of a heterogeneous polymer sample at 3900 points using a visible laser with a wavelength of 532 nanometers. The recorded hyperspectral images were then processed using artificial intelligence and machine learning statistical methods to extract a molecular map of the spatial distribution of polymers in the sample.