DISCUSSION

Methods to measure flow based solely on the washout portion of contrast densograms must use high frame acquisition rates and the implementation of mathematical corrections to avoid errors caused by (1 ) perturbation in the tail- ends of the curves, which can be caused by recirculation, and (2) the delaying effects of contrast material itself. In addition such methods require precise volumes of contrast material. As discussed above the main problem with these methods is the assumption of a constant flow rate, whereas the blood flow in arteries is pulsatile, and this introduces an error in the estimation of blood flow.

Several investigators have attempted to determine mean blood flow rates from time-density curves using transport time methods. For steady-flow conditions, accurate flow rates have been calculated by this technique.

However, for pulsatile-flow conditions, the methods based on finding distinct points on the density curve suffer from relatively high inaccuracy, due to both the inaccurate definition of these points and the fact that in pulsatile flow relative times of passage of contrast between two fixed points cannot be used to measure mean flow.

In both indicator dilution and transport time techniques, flow is assumed to be constant during the measurement period and mean flow is determined. However, arterial blood flow is not constant, and significant errors occur in the mean flow determination, by both techniques, when the flow is highly pulsatile.

To overcome this problem, a new strategy has recently arisen for the measurement of the mean and instantaneous pulsatile blood flow using distance- density curves, as opposed to time-density curves, obtained from digital X-ray angiograms.

In principle the new approach should improve the accuracy of the measurement of pulsatile blood flow waveforms. This is due to the fact that the shape of the distance-density curves appears to be relatively consistent over time, whereas there are much greater changes in the shape of the time-density curves as the bolus proceeds through the vessel under pulsatile flow conditions (see figs 4.2 and 4.3). However, the current techniques used to analyse the distance-density curves are prone to error and particularly sensitive to temporal and spatial variations in background densities.

We have proposed a new technique (Seifalian et al 1988b; 1988c; 1989) for deriving pulsatile blood flow by analysis of time sequences of profiles of contrast material concentration along the vessel. The development and validation of this technique and initial clinical application forms a major part of the work presented in this thesis.

Our X-ray angiographic technique estimates vascular blood flow at the time of injection as the contrast material passes through the vessel(s). Many factors affect blood flow readings in a single subject such as recent food uptake, exercise, anxiety, drug therapy, and many others. Without repeat studies,

therefore, our method provides no information on the variation of flow over a period of time. Subject to X-ray and contrast dose limitations our X-ray angiographic technique is repeatable within a single investigation, as long as further injection of the contrast material does not have effects on blood flow. Investigation at another time will require repeat catheterisation with all the risks that will entail.

Lumen cross sectional area measurements are required to convert velocity blood flow measurements to volume blood flow. Quantitative estimation of cross- sectional area commonly incorporates biplanar X-ray angiograms and assumes circular or elliptical cross-section for measurement of vessel cross-sectional area. As described, various investigators have pointed out that these algorithms may not evaluate adequately the severity of generally eccentric cross-sectional stenoses.

An alternative approach to geometric analysis is the densitometric technique which has usually been applied to the estimation of relative cross-sectional area. This technique uses image intensity information for cross-sectional area measurements. Quantification of the amount of contrast material within a lumen by densitometry provides a measure of cross-sectional area that is independent of lumen geometry. Although accurate relative measurements of vessel narrowing have been demonstrated, it appears that absolute measurements require rigorous validation and correction for the predominant sources of densitometric error (Hawkes et al 1988a). In addition spatial calibration, (i.e. estimation of magnification), is necessary to calculate absolute object dimensions from the radiographic image, and may also be subject to measurement error.

In order to overcome problems with conventional densitometric techniques of measurement of cross-sectional area, Colchester (1984) fitted a semi-ellipse model to the TDP and then computed the cross-sectional area as described above. Initial results assuming that the blood vessel lies parallel to the imaging plane were encouraging and our group later overcame the problem of analysing tortuous vessels using 3D reconstructions of the vascular tree from two X-ray views (Hawkes et al 1988a).

In conclusion, it is clear that the densitometric technique provides potentially more accurate results for measuring the cross-sectional area of normal blood vessels and the only possible solution for measuring blood vessels of eccentric or irregular cross-section in which no geometric assumption about the cross- sectional shape can be made.

In this thesis the technique described by Hawkes et al (1988a) has been used to compute cross-sectional area, which is used to convert flow velocity estimates into volume flow rates.

CHAPTER 5