Development of an implantable blood flow and pressure monitor for pulmonary hypertension : new finite element modelling based conductance catheter techniques for measuring blood flow

Pulmonary hypertension patients could greatly benefit from use of an implantable blood flow and pressure monitor that can be used to derive pulmonary resistance. This thesis concentrates on the development of a suitable implantable sensor for measuring flow. Various methods for measuring flow are assessed. Conductance catheter techniques are proposed as some of the most promising, but suffer from limited accuracy. The main chapters of the thesis concentrate on the development of more accurate conductance catheter techniques. Realistic three-dimensional dynamic models of the heart, developed from MRI and tagged MRI scans, were used with finite element analysis to simulate the electric field arising from the conductance catheter in the heart. Results show that catheter movement and tissue impedance changes give non-negligible error using existing methods for calculating volume from impedance measurements. A new method for calculating volume from impedance measurements was developed that corrects for these and other errors. The new method finds the inverse mapping of a parameterised numerical model. It also allows use of additional data available from the conductance catheter technique, including data from different current source configurations and frequencies. The new technique was tested in a simplified model of the heart. Derivatives of the catheter measurements with respect to the model parameters were determined at one parameter configuration and formed into vectors. Derivative angles were then found between the vectors, providing a method for assessing the separability of the different model parameters. In addition they did not require creation of the inverse mapping. Two different types of techniques for creating the inverse mapping were tested. These were a new technique based on a distance metric and optimisation, and traditional artificial intelligence techniques. Even in the presence of no measurement error, limitations of traditional artificial intelligence techniques prevented successful application. The new technique performed to the order of accuracy of the numerical model on the same amount of data the traditional artificial intelligence techniques were trained on. The new technique was further developed to compensate for measurement error, and resulted in a further increase in accuracy. This thesis has developed and tested new conductance catheter methods, and has moved us a step closer to enabling practical development of an implantable blood flow and pressure monitor.