Author ORCID Identifier
0000-0001-5671-3163
Date of Graduation
8-2023
Document Type
Dissertation (PhD)
Program Affiliation
Medical Physics
Degree Name
Doctor of Philosophy (PhD)
Advisor/Committee Chair
Mark D Pagel
Committee Member
Jingfei Ma
Committee Member
Ken-Pin Hwang
Committee Member
R Jason Stafford
Committee Member
Steven Millward
Abstract
Both tumor acidosis and tumor hypoxia are characteristics commonly found in the microenvironment of solid malignant tumors. Accurate characterization of the two phenomena could provide important information to clinicians for devising suitable treatment plans. Tumor acidosis and hypoxia are closely linked to each other. Tumor acidosis is caused by the inclination of cancer cells towards anaerobic respiration, and one of the main contributing factors for the avoidance of aerobic respiration is tumor hypoxia. Both phenomena can indicate the high metastatic potential of cancers and can also cause resistance of the cancer systems against anti-cancer therapies. Therefore, developing novel molecular imaging techniques is much needed for increasing the accuracy and the precision of the measurement of tumor acidosis and tumor hypoxia. These new molecular imaging methodologies will assist in achieving a better understanding of the cancer microenvironment and improving the quality of clinical care that cancer patients receive.
In this dissertation, I present new methodologies for analyzing data from acidoCEST MRI, which expand the capability of acidoCEST MRI in producing accurate and precise pH measurements. I also present results from a small animal study where electron paramagnetic resonance imaging (EPRI) oximetry was employed to study the oxygenation states of tumor systems. Specifically, in chapter 2 of this dissertation, I present results from, to the best of our knowledge, the first study in which machine learning models were trained with acidoCEST MRI data to accurately and precisely predict the pH levels of iopamidol chemical solutions. The results from this study show that machine learning is a powerful method for analyzing acidoCEST MRI data in both pH classification and pH regression, although the random forest model achieves superior performance in pH regression than the LASSO model. In chapter 3, I optimized the Bloch fitting method and showed that the Bloch fitting algorithm fits for pH levels effectively both from phantom solutions and from in vivo tumor systems. The results demonstrate that no supplementary MR information is needed for the Bloch fitting process and adding potentially inaccurate supplementary MR information can be detrimental and reduce the accuracy of the fitting results. In chapter 4, I use EPRI oximetry to study the hypoxia conditions of three types of tumor models and demonstrate that a new biomarker that measures changes in ΔpO2 can be used to predict the early responses of cancers to radiation therapy as soon as 24 hours after the irradiation process is completed. The results from this study also demonstrate the importance of evaluating the oxygenation state of the cancer in each individual patient, as hypoxia conditions for the same tumor phenotype can vary significantly across subjects. The variation in intratumoral oxygenation can directly affect the efficacy of anti-cancer therapies.
Keywords
MRI, CEST MRI, EPRI, tumor acidosis, tumor hypoxia, molecular imaging, medical physics, machine learning, data modeling