Rhode, KS; (2006) The measurement of arterial blood flow from dynamic digital X-ray images. Doctoral thesis , UCL (University College London).

Preview

Text Rhode_thesis.Redacted.pdf Download (37MB) | Preview

Abstract

This thesis contributes to knowledge by describing three new methods to measure volumetric blood flow waveforms from dynamic digital x-ray images of arteries following the injection of iodine-based contrast material. Three novel approaches were developed to overcome the main limitation of image noise sensitivity of the leading existing methods. The first method was a variation of the concentration-distance curve correlation approach and was termed the polynomial approximation algorithm (PA algorithm). This method modeled concentration-distance curves using polynomial functions prior to determining the optimal shifts between adjacent curves. The second method used a blood flow waveform shape model based on principal component analysis to constrain the blood flow estimates determined by the concentration-distance curve correlation approach. This was termed the model based algorithm (MB algorithm). The third method computed blood flow by performing a weighted average of optical flow estimates along the target vessel with weighting based on the magnitude of the spatial derivative of contrast material concentration. This was termed the weighted optical flow algorithm (OP algorithm). The properties of the OP algorithm were initially investigated using a computer simulation of pulsatile blood flow. Following this, all three algorithms were validated using dynamic x-ray images of progressively more challenging simulated arterial vessels that formed part of a pulsatile blood flow circuit. The x-ray angiographic image data were analysed using a newly developed integrated software package. Flow waveforms were extracted using all three algorithms and compared to simultaneous gold standard recordings from an electromagnetic flow meter (EMF) for a range of different flow rates. The results from the flow circuit experiments showed that the MB and OP algorithms produced flow measurements with low measurement variability and high measurement linearity when compared to EMF values. Furthermore, these algorithms out-performed the basic concentration-distance correlation method and were more robust to image noise and reducing vessel segment length. The PA algorithm showed only minor improvements in performance over the basic approach.

Download activity - last month

Export as XMLCSVJSON

Download activity - last 12 months

Export as CSVJSONXML

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item