|M.Sc Student||Nissim Ronen|
|Subject||Three-Dimensional Structural Measurement of Invading|
Metastatic Cells and their Significance to
Potential Malignancy Classification
|Department||Department of Biomedical Engineering||Supervisor||ASSOCIATE PROF. Daphne Weihs|
Metastasis formation is the major cause of mortality in cancer patients and includes tumor cell relocation to distant organs. A metastatic cell invades through other cells and extracellular matrix by biochemical attachment and mechanical force application. Force is used to move on or through a 2- or 3-dimensional (3D) environment, respectively, or to penetrate a 2D substrate. In previous experiments in the lab, metastatic cells were shown to apply normal force to an impenetrable gel, which led to gel indentation; in contrast, benign cells did not indent the gels. Cells typically apply force through the acto-myosin network, which is mechanically connected to the nucleus. In this work, we focus on identifying the cell elements involved with gel indentation by the cancer cells. Using confocal images of cells indenting the gels, with stains of the nucleus, actin, microtubules, and the membrane, we develop a module to automatically analyze the 3D images, provide quantitative results on relative location of cell elements, sizes of elements, and more. Those results allow us to identify quantitative parameters that differentiate between the benign and cancer cells. This thesis describes the image and data analysis performed for each individual set of confocal images. The confocal images, of fixed and stained cells, include the gel, the cell nucleus, cell membrane and cytoskeletal elements, i.e. the scene; each stain appears in a different (wavelength) channel of the confocal images. We utilize various image extraction and quantification methods to determine the geometry, and relative locations of the cell elements, in order to reveal their potential role in force application. The data images contain an inherent resolution problem in which intensity graining in the images are formed. Thus, a kernel density estimation based pre-process is performed for the stack of each channel, along with channel de-correlation correction and speckle removal. Each pre-processed channel was subsequently segmented and its representative edge 2D surface extracted using a custom made algorithm. Various geometrical measurements of the cell elements location relative to the gel surface were extracted and we evaluated differences that could relate to the metastatic potential (MP) of the cells.
We show that only the cancer cells significantly indent the elastic substrate and the mechanism of indentation may differ between the low and high MP cancer cells. The high MP cells differ from low MP cells and benign in several evaluated parameters. First they displayed a larger indentation depth and lower cell height above the gel. Second, their nucleus indent below gel level and it is significantly less concentric to the cell. All those parameters are in accordance with their highly aggressiveness level. Our results suggest a mechanism with combined roles for both the nucleus and the cytoskeleton that depends on cell aggressiveness. Furthermore, we show that combining indentation related geometrical features, such as the nucleus indentation, along with the nucleus to cell centricity-based measurements offer a feature set which could be a significant factor in a potential cell-population classification process, and should further be explored using extensive statistical and clinical measurements data.