PROFILING EFFUSION CELLS BY QUANTITATIVE ANALYSIS OF MORPHOLOGY AND DIFFRACTION IMAGING PATTERNS

ECU Author/Contributor (non-ECU co-authors, if there are any, appear on document)

Safaa Al-Qaysi (Creator)

Institution

East Carolina University (ECU )

Web Site: http://www.ecu.edu/lib/

Abstract: Profiling cells in human pleural and peritoneal effusion (PPE) samples is an essential task for cytological diagnosis of cancers and patient management. Conventional PPE cytology of a PPE sample is labor intensive and its efficacy depends heavily on the experience of trained specialists in addition to low sensitivity. In this dissertation research project , we focused our effort on application of a label-free method of polarization diffraction imaging flow cytometry (p-DIFC) for quantitative profiling of cell morphology in PPE samples and their correlations to the texture features of cross-polarized diffraction image (p-DI) pairs. To establish the morphology implications of the measured (p-DI) pairs , the 3D structures of PPE cells were measured by using confocal microscopy and quantified with 27 parameters for characterization and analysis of the cellular structures by the conventional fluorescent imaging method. Furthermore , realistic optical cell models (OCM) have been developed as virtual PPE cells and used for accurate simulation of diffraction imaging process to obtain calculated p-DI pairs. This approach allows us to correlate p-DI texture feature parameters quantified by the gray-level co-occurrence matrix (GLCM) algorithm and 3D morphology parameters and investigate various approaches of morphology based cell classification. Clustering algorithms of hierarchical clustering (HC) and Gaussian mixture model (GMM) have been investigated to develop a robust classification method for profiling of the PPE cells' morphological features by the (GLCM) parameters of p-DI pairs. Correlations between the morphological feature parameters and p-DI feature parameters of the imaged PPE cells have been analyzed to gain insights on the morphology implications of image texture patterns and GLCM parameters of the measured p-DI pair data acquired from live and unstained PPE cells of patients of lung and ovarian cancers. Through this dissertation study we have utilized and developed a suite of image processing and analysis tools and obtained results that demonstrate the strong capability of the p-DIFC method to yield big data for profiling PPE cells acquired from cancer patients and the potential to detect malignant PPE cells in the future.

Additional Information

Publication

Dissertation

Language: English

Date: 2019

Keywords

Diffraction imaging, Light scattering, Image analysis, Effusion cells

Subjects

Email this document to