Myocardial fibrosis (MF) underlies several cardiovascular conditions and is difficult to identify with standard histologic techniques. MF is characterized by extensive extracellular matrix remodeling with aberrant deposition of type I and type III collagens in response to various pathological conditions. These lead to diastolic or systolic dysfunction, impairment of intercellular signaling and contractile coordination ultimately leading to heart failure. Despite its prominent role, knowledge gaps exist in detection, quantification and understanding the mechanism of fibrosis, because of lack of suitable histologic techniques. Here we present a novel approach combining discrete frequency infrared imaging and machine-learning to develop a label-free, rapid, non-destructive workflow fully compatible with intraoperative pathology. Our method exploits the unique sensitivity and specificity of infrared spectroscopy towards collagen (area under receiver operator characteristic curve 0.99). Subsequently we train a conditional Generative Adversarial Network (c-GAN) to computationally generate super-resolved virtual picrosirius red stain (the gold standard in collagen specific stain) from IR data. We demonstrate the applicability of our method for three different classes of patients: 1) nonpatholgic (control), 2) heart transplant, 3) patients undergoing implantation of ventricular assist device. Our virtual stains show excellent visual and quantitative agreement with chemically stained samples for three different morphological subtypes of MF namely the replacement, interstitial and perivascular fibrosis. The average infrared spectra of the different subtypes show reproducible differences suggesting differences in the underlying composition of the collagen fibers arising out of either the subtype of collagen or post-translational modifications. Overall, our results demonstrate a clinically translatable, objective and quantitative method for diagnosis and for obtaining mechanistic insights into the pathophysiology of MF.

Application of infrared spectroscopic imaging for label-free detection and quantification of myocardial fibrosis

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.