Channel characterisation and modelling for transcranial Doppler ultrasound.
Weir, Alexander James
MetadataShow full item record
The detection of micro-embolic signals (MES) is a mature application of transcranial Doppler (TCD) ultrasound. It involves the identification of abnormally highpitched signals within the arterial waveform as a method of diagnosis and prediction of embolic complications in stroke patients. More recently, algorithms have been developed to help characterise and classify MES using advanced signal processing techniques. These advances aim to improve our understanding of the causes of cereberovascular disease, helping to target the most appropriate interventions and quantifying the risk to patients of further stroke events. However, there are a number of limitations with current TCD systems which reduce their effectiveness. In particular, improvements in our understanding of the scattering effects in TCD ultrasound propagation channels will benefit our ability to develop algorithms that more robustly and reliably identify the consistency and material make-up of MES. This thesis explores TCD propagation channels in three related research areas. Firstly, a method of characterising TCD ultrasound propagation channels is proposed. Isotropic and non-isotropic three dimensional space (3-D) spherical scattering channel models are described in terms of theoretical reference models, simulation models, and sum of sinusoids (SoS) simulators, allowing the statistical properties to be analysed and reported. Secondly, a TCD ultrasound medical blood flow phantom is described. The phantom, designed to replicate blood flow in the middle cerebral arteries (MCA) for TCD ultrasound studies, is discussed in terms of material selection, physical construction and acoustic characteristics, including acoustic velocity, attenuation and backscatter coefficients. Finally, verification analysis is performed on the non-isotropic models against firstly, the blood flow phantom, and secondly, a patient recordings database. This analysis expands on areas of agreement and disagreement before assessing the usefulness of the models and describing their potential to improve signal processing approaches for detection of MES. The proposed non-isotropic channel reference model, simulation model, SoS simulator, and blood flow phantom are expected to contribute to improvements in the design, testing, and performance evaluation of future TCD ultrasound systems.