Endoscopic Fluorescence Lifetime Imaging Microscopy (FLIM) Images of Aortic Plaque: An Automated Classification Method

摘要

The objective of this study was to develop an automated algorithm which uses fluorescence lifetime imaging microscopy (FLIM) images of human aortic atherosclerotic plaque to provide quantitative and spatial information regarding compositional features related to plaque vulnerability such as collagen degradation, lipid accumulation, and macrophage infiltration. Images were acquired through a flexible fiber imaging bundle with intravascular potential at two wavelength bands optimal to recognizing markers of vulnerability: F_(377): 377/55 nm and F_(460): 460/50 nm (center wavelength/bandwidth). A classification method implementing principal components analysis and linear discriminant analysis to correlate FLIM data sets with histopathology was validated on a training set and then used to classify a validation set of FLIM images. The output of this algorithm was a false-color image with each pixel color coded to represent the chemical composition of the sample. Surface areas occupied by elastin, collagen, and lipid components were then calculated and used to define the vulnerability of each imaged location. Four groups were defined: early lesion, stable, mildly vulnerable and extremely vulnerable. Each imaged location was categorized in one of the groups based on histopathology and classification results; sensitivities (SE) and specificities (SP) were calculated (SE %/SP %): early lesion: 95/96, stable: 71/97, mildly vulnerable: 75/94, and extremely vulnerable: 100/93. The capability of this algorithm to use FLIM images to quickly determine the chemical composition of atherosclerotic plaque, particularly related to vulnerability, further enhances the potential of this system for implementation as an intravascular diagnostic modality.