Raman spectral classification of atherosclerosis using neural networks and discriminant analysis

Raman spectroscopy is a powerful non-destructive technique and has a high potential for in vivo diagnosis applications of atherosclerotic plaques in human arteries. For such real time clinical applications, a rapid collection and analysis of the data is needed. One of the major problems with rapid data collection is that the noise generated by the detector (even with one of the most advanced versions) has the same level as the Raman signal from the tissue which makes the analysis difficult. In this paper, different processing techniques for compressing the spectrum vector collected with very short time scales (∼msec) and its rapid classification methods were analyzed. The accomplished results demonstrated that the classification error was smaller than 5%, even with the data collection times as low as 50 msec, when the wavelet transformation was utilized to compress the input vector and the classification methods based on either neural network or discriminant analysis were applied.