Date of Graduation

12-2020

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Engineering (PhD)

Degree Level

Graduate

Department

Biomedical Engineering

Advisor

Kyle P. Quinn

Committee Member

Colin D. Heyes

Second Committee Member

Timothy J. Muldoon

Third Committee Member

Narasimhan Rajaram

Keywords

autofluorescence, deep learning, in vivo, microscopy, redox metabolism, wound healing

Abstract

Non-healing ulcerative wounds that occur frequently in diseases such as diabetes are challenging to diagnose and treat due to numerous possible etiologies and the variable efficacy of wound care products. With advanced age, skin wound healing is often delayed, leaving elderly patients at high risk for developing these chronic injuries. As it is challenging to discriminate age-related delays from disease-related chronicity, there is a critical need to develop new quantitative biomarkers that are sensitive to wound status. Multiphoton microscopy (MPM) techniques are well-suited for 3D imaging of epithelia and are capable of non-invasively detecting metabolic cofactors (NADH and FAD) without exogenous stains. Unfortunately, data collection requires time-consuming, manual segmentation of wound images. Convolutional neural networks (CNNs) are specialized algorithms sensitive to subtle changes in tissue features and can provide accurate, consistent image segmentation in only a few seconds. The objective of this study was to evaluate the utility of label-free MPM for characterizing wound healing dynamics in vivo and identifying potential differences between healthy and pathologically delayed wounds. Using full-thickness, excisional skin wounds produced on the dorsum of living mice, an optical redox ratio of FAD/(NADH+FAD) autofluorescence was isolated and measured alongside NADH fluorescence lifetime (FLIM) images to provide 3D maps of wound metabolism over 10 days. In vivo MPM enabled the spatial isolation of epithelial keratinocytes where changes in redox ratio and NADH lifetime were observed in murine models of type I diabetes and advanced age. Specifically, keratinocytes in diabetic mice remain in a proliferative state longer than control wounds indicated by a lower redox ratio. Aged mice exhibited a higher epidermal redox ratio than young mice, suggesting a decreased proliferative capacity. NADH FLIM images also revealed temporal changes in the epidermal NADH lifetime, which correlated with the optical redox ratio during proliferation. Finally, a series of CNNs were trained to segment 3D MPM images of wounds to automatically quantify wound biomarkers such as the optical redox ratio. This work demonstrates label-free MPM can provide non-invasive metabolic biomarkers sensitive to temporal and conditional changes in skin wound healing and an automated analysis pipeline based on CNNs can assist in the rapid detection of impaired healing.

Available for download on Wednesday, January 25, 2023

Share

COinS