Author:

Abstract:

This doctoral thesis addresses the issue of geometrical calibration of CT instruments. The term 'calibration' has a well-defined meaning in the field of metrology: measurement of a quantity by comparison to a traceable reference and assessment of uncertainty in that comparison. Therefore, instrument geometrical calibration refers to the measurement of the instrument geometry, defined by a set of geometrical parameters, by comparison to a traceable reference and assessment of uncertainty in the measured values. The doctoral journey has followed a relatively common path in the quest for developing standardized methods to measure the geometry of a measuring instrument. After realizing that standardized methods don't exist by way of a literature review, the sensitivity of measurements to various geometrical error sources is determined. Sensitivity analysis allows us to determine which error sources are negligible so that they may be put on the back burner of any research endeavor. Dedicated procedures for the measurement of the CT instrument geometry by comparison to a traceable reference are developed and applied to simulated data. Implementation on simulated data provides us with ground truth, which we can use to evaluate the performance of the test procedures and adapt as necessary. The geometrical measurement procedure is applied to an experimental instrument. While ground truth is not available for the experimental implementation, i.e. we don't know the true instrument geometry, the efficacy of the developed procedures is validated by observing considerable reductions in measurement errors after compensation of the measured geometrical misalignments by instrument adjustment. The development of a Monte Carlo framework for assessing uncertainty in CT instrument geometrical parameters solved by minimization is discussed. However, application of the Monte Carlo framework is currently limited due to bias in the input data and coupling of the solvable parameters. The Monte Carlo framework and the limitations outlined provide a path for such future research endeavors. Furthermore, the Monte Carlo framework proposed here can be applied to evaluate uncertainty in any optimization-based measurement procedure.