Estimating Viscoelastic and Failure Properties of Thoracic Aorta Using Near Infrared Spectroscopy

Abstract

The thoracic aorta is the proximal segment of the aorta, comprising the ascending aorta (AA), aortic arch (AAR), and descending aorta (DA). Thoracic aorta diseases, such as aneurysms, are highly fatal if left untreated. The initiation and progression of those diseases are associated with alterations in mechanical properties. The mechanical characterization of the thoracic aorta is fundamental for understanding the mechanobiology of aortic disease and for the development of biomechanics-based rupture risk stratification strategies. Currently available characterization methods are ex vivo and destructive, which limit their clinical applicability. Consequently, non-destructive and in vivo characterization techniques are highly required for accurate mechanical properties assessment.

This thesis explored the suitability of near-infrared spectroscopy (NIRS) augmented with machine learning as a potential non-destructive assessment tool. Thoracic aorta samples harvested from two porcine aortae were prepared along the longitudinal direction. Near-infrared (NIR) spectra were acquired from six measurement locations on each sample. Following spectral acquisition, each sample underwent uniaxial tensile testing, from which the viscoelastic and failure properties were extracted. Random forest and gradient boosting regression models were then trained to correlate the NIR spectra with the extracted properties.

Keywords:

Thoracic aorta, biomechanical properties, Near-infrared spectroscopy, non-destructive assessment, machine learning, viscoelastic properties, failure properties, aortic aneurysm.