| Summary: | The study of the effects of damaging agents on living cells is of importance both for the assessment of the potential consequences of exposure, and for the advancement of medical technologies concerned with the treatment of cancer and other diseases. Such agents, however, are commonly studied at doses and levels of effect much higher than those seen in environmental or even medical exposures. Studies of cell survival after treatment with ionizing radiation, for instance, generally involve the measurement of dose-response over several decades of cell kill on a logarithmic scale. In contrast, cell survival rates for a typical clinical treatment dose are on the order of 50%. Survival measurements in this first decade of cell kill require that the exact fate of thousands of cells is determined in a single experiment. This is beyond human capabilities. Measurements at these levels of effect therefore require the development of a rapid, automated system of cell detection, characterization, and follow-up.
The primary aim of this thesis was to test the hypothesis that the algorithms necessary for generating survival data with such an automated system can be developed to standards of accuracy comparable to those of an experienced human observer. Using an image cytometry device specifically designed for the detection and analysis of live, unstained cells, automated scanning procedures were optimized for selected cell lines, and means of maintaining appropriate focus levels during a scan were devised. Algorithms to distinguish cells from other objects detected in the flask were also developed. These performed with comparable accuracy, but at greater speed, than could be achieved by a human observer.
Because the endpoint used in survival measurements is the ability of treated cells to proliferate to form colonies, the hypothesis that automated methods of assaying colony formation could be developed was also tested. Using rapidly collected, low resolution image data obtained at locations in the tissue culture vessel where individual cells had been detected on the day of treatment, it was found that the survival status at 70-90% of these locations could be determined automatically. Manual assessments were required at the remaining locations.
The final objective of this thesis was to use the methods developed for automated sample selection and survival assessment to examine a question relevant to both radiotherapeutic applications and the understanding of mechanisms of radiation action. In particular, the Theory of Dual Radiation Action (Kellerer and Rossi, 1972), which predicts that the relative biological effectiveness (RBE) of different modalities of sparsely ionizing radiations may change dramatically with decreasing dose, was tested experimentally. Using two different mammalian cell lines, the experimental data in general rejected the predictions of this theory, although a modified theory, developed by the same authors, could accommodate the results. In summary, the data showed that there may be a slight dose dependence in the RBE of the radiations tested. Specifically, the RBE of low energy X-rays relative to ⁶⁰Co γ-rays was found to increase slightly with decreasing dose. The measured RBE's in the zero-dose limit were ~1.4 for 55 kVp X-rays and ~1.1-1.2 for 250 kV[subscript]p X-rays. High energy (11 MeV) electrons, on the other hand, showed a small decrease in RBE relative to ⁶⁰Co γ-rays as the dose approached zero, having a zero-dose limit of ~0.95. All of the aforementioned radiation modalities had an RBE of 1.0-1.1 at high doses (10 Gy or more). === Science, Faculty of === Physics and Astronomy, Department of === Graduate
|
|---|
