This item is under embargo and not available online per the author's request. For access information, please visit http://libanswers.wustl.edu/faq/5640.
Date of Award
Doctor of Philosophy (PhD)
Being the routine clinical practice for most cancer types, assessing tumor histopathology is critical for cancer diagnosis and prognosis. Histological reviews by clinical pathologist based on the tissue from biopsy or surgical resection remain the only definitive diagnosis of tumor pathologies. However, biopsy or surgical resection is invasive with potential adverse side-effects, making it urgent to develop noninvasive imaging techniques for assessing tumor histopathology. Diffusion MRI was proved to be sensitive to cancer detection in several types of cancer. Yet, current diffusion MRI methods are not specific enough to assess tumor histopathology, especially for cancers like glioblastoma (GBM) and prostate cancer, most with complicated tumor micro-environment. To address this challenge, we employ a novel Diffusion Histology Imaging (DHI) approach, combining diffusion basis spectrum imaging (DBSI) and machine learning/deep learning, to accurately and non-invasively assess tumor histopathology. We apply DHI in imaging patients with GBM to reveal potential viable tumor and necrosis regions that current clinical imaging gold is not able to detect. For validation, we examined twenty surgical resection specimens from thirteen GBM patients and demonstrated that DBSI-derived restricted isotropic diffusion fraction significantly correlated with GBM tumor cellularity. The results further indicated that DHI predicted high cellularity tumor, tumor necrosis, and tumor infiltration with accuracy rate of respectively 91.9%, 93.7%, and 87.8%. It was suggested that DHI might serve as a favorable alternative to current neuroimaging techniques in guiding biopsy or surgery as well as monitoring therapeutic response in the treatment of glioblastomas. Similarly, we applied DHI on prostatectomy specimen and prostate cancer patients, and it was highly accurate not only in detecting prostate cancer from other benign prostatic histology or structures, but also in classifying various prostate cancer grades (grade 1: 88%; grade 2: 94%; grade 3: 92%; grade 4: 88%; grade 5: 95%). We demonstrated that through evaluating and profiling various histopathological structures in prostate cancer, DHI could increase accuracy of tumor detection, staging and grading.
Chair and Committee
Joseph J. Ackerman Sheng-Kwei Song
Joseph J. Ackerman, Sheng-Kwei Song, Sophia E. Hayes, Dewey Holten,
Ye, Zezhong, "Noninvasive Histopathological Imaging of Brain and Prostate Cancer" (2019). Arts & Sciences Electronic Theses and Dissertations. 1968.
Available for download on Tuesday, August 15, 2119
Analytical, Diagnostic and Therapeutic Techniques and Equipment Commons, Artificial Intelligence and Robotics Commons, Bioimaging and Biomedical Optics Commons, Physical Chemistry Commons, Radiology Commons