Aspects of modelling in medicine
Thesis DisciplineElectrical Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
Three applications of models in medicine are considered. A computer-aided learning program for teaching the dynamics of uptake and distribution of the inhalational anaesthetic halothane is described. The program is based on a compartmental model which simulates the action of halothane on ventilation and on the cardiovascular system. The program presents the model to the student in four forms: one with no changes in circulation or respiration, one with the cardiovascular effects of halothane included, one with respiratory effects only, and one with both of these effects combined. The student can study the importance of halothane's influence on respiration and blood circulation by comparing results from simulations on different models. The simulation is presented as graphs continuously displayed on an alphanumeric visual display terminal. Interaction with the program is possible at all times to change the simulation speed, the variables being graphed, the halothane fraction, and the fresh gas flow. A pilot evaluation of the program as it is used during a tutorial shows a highly significant improvement in the students’ answers after the tutorial using a 'before and after' questionnaire. The students showed an understanding of the program's display and model limitations. This encourages the further use of the program. The images obtained by Computed Tomography are degraded if there is any body motion while the projections are being measured. This blurring is assessed by considering motions of individual points within body cross-sections. An effective radius of body movement is defined and a statistical description of the blurring is developed. Both continuous and discrete movements are investigated. It is found that the resolution limit set by the blurring is almost always twice the effective radius, although the blurring can sometimes be negligible if the movement occurs spasmodically during the measurement of the projections. In some situations of technical and scientific interest, only small numbers of projections are available. For these, conventional reconstruction techniques produce unacceptable images. Several methods for increasing the number of projections by interpolating between given projections are introduced and compared. It is found that schemes which employ a priori information give images which are superior to those which do not utilise such information. The general structure, rather than the specific details (e.g. whether or not the underlying functional is of maximum entropy type) of an iterative scheme determines its usefulness as an interpolation method. It is shown how the convergence rates of such iterative schemes can be accelerated.