NWU-RSIT: An integrated graphical user interface for biomedical Raman spectral imaging with both univariate and multivariate modules

When consecutively acquiring Raman spectra at a given number, Raman imaging could be applied to extract the compositional and structural information from the selected sampling region. By compacting both univariate and multivariate algorithms, an integrated Raman spectral imaging toolbox (NWU-RSIT) was developed for spectral analysis, image reconstruction, and feature recognition. Using an example hyperspectral dataset from a single living mouse osteosarcoma cell, univariate imaging method reveals spatial distributions of some specific molecular vibration modes; multivariate algorithms, including principal component analysis (PCA), hierarchical clustering analysis (HCA), k-means clustering analysis (KCA), vertex component analysis (VCA), and N-FINDR algorithm, were realized to extract more detailed constitution information from the acquired hyperspectral dataset. After comparing and evaluating the system performance of NWU-RSIT, it will make Raman spectral imaging more accessible to all related fields, especially for the biomedical studies.

Automated summary

An integrated Raman spectral imaging toolbox (NWU-RSIT) was developed for spectral analysis, image reconstruction, and feature recognition by combining univariate and multivariate algorithms. Different algorithms, such as PCA, HCA, KCA, VCA, and N-FINDR, were used to extract detailed composition information from hyperspectral dataset, making Raman spectral imaging more accessible for biomedical studies.