Label-free discrimination of lung cancer cells through mueller matrix decomposition of diffuse reflectance imaging

Abstract

Abstract In this article, we explore the potential of an original label-free Near-Infrared (NIR) imaging technique, based on Mueller Matrix decomposition reflectance, for efficient detection and classification of histopathological samples of lung cancer cells. Experimental results were acquired, processed, and analyzed by means of an accurate, fully-automated, auto-calibrated liquid-crystal NIR polarimetric imaging system, developed for real-time Mueller matrix analysis and optical characterization of target media. The polarimetric Figure-of-Merits (FOMs), estimated using Mueller matrix decomposition, as well as the statistics associated with the sixteen Mueller matrix elements of each lung cell sample indicate that enhanced discrimination among the samples can be achieved. Similarly, polarimetric Exploratory Data Analysis (pEDA), based on histograms obtained from diffuse reflectance polarimetric signals, has been used to determine if aberrations and/or changes in the spread of the histogram between different stages of lung cancer can be proved effective biomarkers for its progression and also discrimination among different lung pathologies. The outcome of this study indicates that Mueller matrix formalism may be proved extremely useful in discriminating among healthy and malignant lung cells as well as differentiating among the different types of malignancies with high accuracy. As a result, it may contribute positively to the enhancement and implementation of the operational principles of the Whole Slide Imaging (WSI) field.