Date of Award

1-1-2020

Language

English

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

College/School/Department

Department of Chemistry

Content Description

1 online resource (xxii, 127 pages) : color illustrations.

Dissertation/Thesis Chair

Igor K Lednev

Committee Members

Jeremy Feldblyum, Aidar Gosmanov, Alexander Shekhtman, Jia Sheng

Keywords

Chemometrics, Machine Learning, Medical Diagnostics, Raman Spectroscopy, Vibrational Spectroscopy, Raman spectroscopy, Diagnostic imaging

Subject Categories

Analytical Chemistry

Abstract

Many problems exist within the myriad of currently employed diagnostic techniques. Further, an incredibly wide variety of procedures are used to identify an even greater number of diseases which exist in the world. There is a definite unmet clinical need to improve the diagnostic capabilities of these methods, including improving test sensitivity and specificity, objectivity and definitiveness, and reducing cost and how invasive a test is, with an interest in replacing multiple diagnostic methods with one powerful tool. That is, the development of a singular technique which can accurately and objectively diagnose a wide variety of diseases in a cost-effective and non-invasive manner has the powerful potential to revolutionize the field of medical diagnostics. Raman spectroscopy in combination with chemometrics is proposed here as a method which can satisfy this essential need. First, the method of Raman spectroscopy in combination with chemometrics was applied for investigating several different diseases in an effort to improve methods for diagnosing them. These include Alzheimer’s disease, Duchenne muscular dystrophy, and Celiac disease in separate proof-of-concept studies. In each case, a statistical model was built for the purpose of differentiating Raman spectra collected from biological samples of healthy donors and donors with the disease. The performance of each model was evaluated through external validation to assess the significance and success of the method. In each study, over 95% accuracy was achieved, illustrating the method is robust for accomplishing medical diagnostic applications. Second, deep-ultraviolet resonance Raman spectroscopy (DUVRS) was used to investigate its potential for future medical diagnostic applications. DUVRS is advantageous due to the lack of fluorescence interference which occurs at excitation wavelengths shorter than 250 nm and due to the resonance enhancement of biological materials such as nucleic acids and aromatic amino acids owing to their absorption of light in the same range. DUVRS was used in a proof-of-concept study to probe the spectral differences between cancerous and normal prostate cell lines as well as cancerous and adjacent healthy brain tissue. This preliminary study indicated spectral differences exist between the groups, demonstrating significant potential for additional work to utilize DUVRS for improving methods for diagnosing cancer. DUVRS was further used to understand its applications for monitoring cell growth and differentiation. Four stages of cellular differentiation were investigated, with chemometric analysis showing successful classification of the four stages. These studies open the door for further work which may increase the use of DUVRS for monitoring the growth of cells and their progression toward disease. Lastly, surface enhanced Raman spectroscopy (SERS) was used to expand upon the potential of Raman spectroscopy for medical diagnostics. SERS is a sensing technique where inelastic scattering of light is enhanced due to the adsorption of molecules onto rough metal surfaces, such as silver or gold nanoparticles. The SERS effect can enhance inelastic scattering by factors up to 1010, or larger. SERS was explored here as a method which can accurately detect very low concentrations of biomolecules with high levels of specificity. SERS experiments were performed using gold substrates manufactured by electrochemical deposition; results showed DNA could be detected at the single-molecule level at an ultralow concentration. The successful identification of short DNA strands indicates the potential for the method to be applied for detecting other biologically relevant biomolecules, including RNA, which are known disease biomarkers. Altogether, this research aims to develop and expand the method of Raman spectroscopy for improving medical diagnostic approaches, with a potential to be applied as a universal medical diagnostic technique in the future. Raman spectroscopy with chemometrics was directly applied for diagnosing specific diseases with high levels of accuracy in proof-of-concept studies. Both deep-ultraviolet resonance Raman spectroscopy and surface enhanced Raman spectroscopy were preliminarily investigated for their potential to contribute toward improving medical diagnostic measures. Further research is required to validate the results of these studies in large-scale, statistically significant clinical trials, as well as to test the specificity of the method for diagnosing similar diseases. Based on the results presented herein, the method of Raman spectroscopy in combination with chemometrics shows a strong ability to improve and even revolutionize the field of medical diagnostics as the first minimally invasive, accurate, rapid, and potentially universal medical diagnostic technique.

Share

COinS