Classification of aortic atherosclerotic lesions with time-resolved fluorescence spectroscopy

In this study, we examine the possibility of differentiating between classes of atherosclerotic lesions based on time-resolved fluorescence spectroscopy and we compare the performance of classification schemes that use either the time-resolved spectra or only the intensity spectra. Transient fluorescence emission induced by pulsed nitrogen laser excitation was measured on 87 excised samples of human aorta. The samples were classified histologically using the AHA classification (normal to type VI lesion). Predictor variables derived from the time-resolved spectra included the spectral intensities at 360-510 nm (10 nm interval) and parameters of a biexponential fit of the fluorescence impulse response function. Stepwise discriminant analysis using these predictor variables showed that a few predictor variables (time-dependent decay parameters at 390 nm, spectral intensities at 460 and 490 nm) sufficed to correctly classify 89% of the samples. Excluding the time-dependent decay and using only the spectral intensities, the percentage of correctly classified cases was significantly lower: 51%. These results establish that time-resolved fluorescence spectroscopy markedly improves on the performance of steady-state fluorescence spectroscopy for fine classification of atherosclerotic lesions.