Pixel-level angular quantification of capsular collagen in second harmonic generation microscopy images of encapsulated thyroid nodules.

Abstract

Polarization-resolved second harmonic generation microscopy is used to provide pixel-level angular distribution of collagen in thyroid nodule capsules. The pixel-level angular distribution is combined with textural analysis to quantify the collagen distribution in follicular adenoma (benign) and papillary thyroid carcinoma (malignant). Three second order nonlinear susceptibility tensor elements ratios, the collagen angular distribution and two parameters accounting for the collagen angular dispersion in different sized areas are extracted and corresponding images are computed in a pixel-by-pixel fashion. Subsequently, we show that texture analysis can be performed on these images to detect significant differences between the considered benign and malignant nodule capsules.