This thesis deals with the estimation of blood flow in the heart and larger vessels where control-volume methods are applied using ultrasound Doppler technique. In particular two control-volume techniques were investigated: The proximal isovelocity surface area method, (PISA) and the Surface Integration of Velocity Vectors method, (SIVV).
For PISA, computational fluid dynamics, (CFD) was used for non-stationary flow and non-planar circular geometries where special emphasis was given to the influence from the angle of the valvular leaflets on the proximal surface area. The CFD results were compared with ultrasound measurements, in an in-vitro model with controlled geometry and flow characteristics. Three different valvular geometries were used: planar, reversed cone and funnel. In these idealised CFD and experimental models it was found that there is support to use the hemispherical PISA approach for the geometries investigated provided that the flow is not to high in the reversed cone and funnel case. At high flows the actual proximal geometry should be used instead of an entire hemisphere.
A hydraulic pulsatile model was used in developing a platform with in-house software where the SIVV flows automatically may be calculated from a digitally stored raw data. An antialiasing algorithm was developed to allow for measurement of aliased data in order to increase the dynamic velocity range. The antialiasing algorithm was found to improve the estimation of SIVV flow.
The influence on the flow estimate was investigated with respect to the number of scan-planes using a numerical model and in-vitro and in-vivo model experiments. It was found that a minimum of two scan-planes are needed when flow conditions and geometry is close to circular, otherwise the recommendation is four scan-planes.
A steady state and a pulsatile model was used to evaluate accuracy of the SIVV method more extensively in vitro. SIVV was found to be accurate and repeatable with a slight underestimation in the pulsatile model but within the ±10% range. In the steady state model a strong correlation was found between SIVV and timed flow. However, since discrepancies in regression equations were obtained for different tube diameters further investigation of steady state flows in vessels of small diameter are needed.
An in-vivo model was designed to study the possibility to use the SIVV method to measure cardiac output in a paediatric model in haemodynamically unstable subjects and to investigate what measurement site to use. Epicardial measurements were performed on a series of piglets using two different temporal resolutions. SJVV accuracy was compared with ultrasound transit time flow and was found to be in parity or better than current invasive methods. Inter- and lntraobserver variability was found to be low.
Linköping: Linköpings universitet , 2000. , 48 p.
2000-04-17, Berzeliussalen, Universitetssjukhuset, Linköping, 09:15 (Swedish)