RUcore Resource Object
RUcore Resource Object
PDF-1(4.98 MB)
TitleAnalysis of cancer progression and morphology using computational modeling and image processing techniques
NameNorton, Kerri-Ann (author), Shinbrot, Troy (chair), Bhanot, Gyan (internal member), Axelrod, David (internal member), Barnard, Nicola (outside member), Ganesan, Shridar (outside member), Rutgers University, Graduate School - New Brunswick,
Degree Date2011-05
Date Created2011
SubjectComputational Biology and Molecular Biophysics,
Breast--Cancer--Magnetic resonance imaging,
Breast--Cancer--Epidemiology
DescriptionIn the US, an estimated 560,000 people died from cancer in the year 2009 (Jemal et al., 2009). The goal of this dissertation is to improve the fundamental understanding of cancer morphogenesis and so introduce scientific rigor into the diagnostic and prognostic methods used in the clinical setting. To this end, we focus on defining relationships between the history of growth of a cancerous lesion--which is what reliable prognosis depends on – and the morphology observed at an instant in time – which is what clinicians can observe ‐ through a combination of computational modeling and image processing. We achieve this goal by completing three aims. The first aim of the thesis is to examine the architectural progression of ductal carcinoma in situ of the breast (DCIS) using a two‐dimensional computational model. In this work we have found that the distinct architectural subtypes can result from different cellular features or from precancerous growths with similar cellular features but observed at different time points. The second aim of this thesis is to develop a border detection algorithm for skin lesions collected by dermoscopy. This work has resulted in the production of an automated and experimentally validated computational tool for the discrimination of melanoma, benign melanocytic lesions and non‐meanocytic lesions. The third and final aim is to characterize the fully three‐dimensional (3D) morphology of DCIS. In this work, we developed a 3D reconstruction approach to build 3D representations of DCIS. From this research, we have determined for the first time that there are two distinct 3D architectures of cribriform DCIS that are indistinguishable in 2D cross‐sections. Based on this finding, we propose that 3D reconstructions hold additional clinical information that cannot be accessed from analysis of 2D histological samples.
NotePh.D.
NoteIncludes bibliographical references
Noteby Kerri-Ann Norton
Genretheses
Persistent URLhttp://hdl.rutgers.edu/1782.1/rucore10001600001.ETD.000061395
Languageeng
CollectionGraduate School - New Brunswick Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.