LED based Raman imaging system

• Main Author: Ng, Zong Yi
• Other Authors: Liu Quan
• Format: Final Year Project
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Bioengineering
• Online Access: http://hdl.handle.net/10356/65207
• Institution: Nanyang Technological University

Lasers are often used as an excitation source for Raman spectroscopy due to its capability of emitting monochromatic and high intensity photons. However lasers are expensive and are not always reliable in several cases, such as sample degradation. Therefore broadband light sources, such as Light Emitting Diodes (LEDs) being inexpensive and widely available in all spectral ranges, can provide an alternative to develop a low-cost solution for Raman spectroscopy. However LEDs are low in power and have large spectral bandwidth, thus the use of LEDs as excitation source will result in chromatic dispersion and not only excite the element of interest, which will result in a distorted spectrum. In this project, the distorted Raman spectrum when excited by an LED excitation source were simulated by convoluting the original Raman signal from phantom data with LED output spectrum. Algorithm-based techniques such as FFT deconvolution and narrow-band measurements based on Wiener estimations were utilized to restore the original Raman spectrum. The results suggested that both methods presented were effective in retrieving the peak positions of the original spectrum, however intensity of the spectrum varies across methods. RMSE value were also used to validate the accuracy of the restoration, the RMSE results revealed that averaging the kernel set demonstrated to be the most effective method in FFT deconvolution. While in narrow-band measurements, RMSE varies significantly across different filter combinations. However with the correct selection of filters, narrow-band measurements proved to be an effective method to restore the original spectrum.