Digital processing of Raman spectra for diagnosis of atherosclerosis

Discusses the digital processing methodology utilized to analyze Raman spectral data, with the ultimate aim of developing a rapid and automatic system for atherosclerosis diagnosis. Different types of digital and wavelet transform filters have been studied in order to reduce the CCD detector noise. After calibration, Raman spectra have been processed by an automatic program that classifies the target tissue into pathological or non-pathological using pattern recognition techniques. To validate the diagnosis inferred by the automated system, a collection of 70 spectra from human coronary arteries has been tested and compared with the histological method. The processing time of whole analysis is as small as 10 milliseconds when the program is executed on a processing station based on the ADSP 61061 Sharc digital signal processor.