Signal processing for fast Raman imaging under low SNR condition and its applications in Raman imaging

Imaging in Raman spectroscopy is a valuable tool for analytical chemistry, cell research and so on. At present, low SNR (signal-noise ratio) of the Raman spectral data set and the data processing speed of the imaging process are the two constraints of observing some dynamic change reactions. In this paper, a fast univariate Raman imaging algorithm named CPSA (Characteristic Peak Sparse Approximation) which is based on sparse approximation, least square and Butterworth low pass filter was presented to generate the Raman image under low SNR condition, ant it was proved to be timesaving as well as high imaging quality. In CPSA algorithm, sparse approximation is used to eliminate the noises of the target Raman signals; least square is used to enhance the useful Raman signals; Butterworth low pass filter is used to improve the image quality. Three experiments were conducted and the results were presented. The first experiment was used to validate the reliability of the proposed method, and the left two experiments were its applications in medical and human cancerous cell imaging.