In spite of the novelty of the report and the extensive analysis of the authors on the possible sources of high variability of volumetric flow measurements, it appears to us that there are several points which need further consideration; some of these were analysed in the editorial by Hedges,2 but we would like to highlight two additional questions.
For us, a weak point of the study protocol of Orge and colleagues is their method for the determination of the diameter of the ophthalmic artery. They assume the boundaries of the vessel to be where detected movement starts and ends along the m-mode line; or in other words the vessel boundaries are taken to be the positions where the grey pixels and colour pixels touch on this Doppler image. However, we are doubtful whether this definition is relevant for the required quantitative measurement. The first point is that for small vessels the width of the colour area (indicating blood motion) is unfortunately relatively independent of the true vessel diameter. The width of the superimposed colour area is greatly influenced by the actual technical parameters used in colour Doppler imaging (pulse repetition frequency, lateral dimension of the ultrasound beam, colour priority, motion discriminator setting, colour saturation, brightness, contrast, etc). Our second doubt is that, even on the grey scale part of the image shown by Orge et al in their in Figure 1, no vessel wall is seen, unlike the case for typical images of large vessels like the carotid arteries.
 | Figure 1. CVI flowmetry of the dilated superior ophthalmic vein in a 70 year old patient with direct carotidcavernosus sinus fistula. In the upper left image the thin white line indicates the sonic beam-flow angle, for volume-flow calculation. The functional lumen (more ...) |
We think that because of these difficulties regarding the determination of the vessel wall position, Orge and co-workers overestimate the ophthalmic artery diameter. Their diameter estimate is 2.02 mm on average; but this figure is significantly larger than is suggested by other evidence. During conventional 10 MHz B-scan diagnostic examination, the ophthalmic artery is never visible. However for the dilated ophthalmic vein, in exceptional cases such as in a patient with carotideo-cavernous sinus fistula, or in a small baby in a bout of strenuous crying, the vein is then well outlined with a diameter of 1 mm or above. Thus, we would expect a 2 mm diameter artery to be clearly visible. As we demonstrated some years ago,3,4 patients with a pathologically dilated ophthalmic vein are good candidates for non-invasive volumetric blood flow measurement. We were able to measure volumetric blood flow in the orbit of patients with a high flow fistula (vein diameter around 3–4 mm) using the CVI-Q technique (Fig 1). Possibly future improvements in spatial resolution may resolve this difficulty.
In a vessel like the ophthalmic artery there is a further problem in determination of the average velocity, because the laminar flow in such small vessels causes a very wide velocity variation within the lumen, as can be seen from our Figure 1. In contrast, we note that the colour spectrum in the figure presented by the authors is almost completely uniform and does not show higher speed in the centre of the lumen compared to that close to the vessel wall. This might imply a relative insensitivity of velocity discrimination within a small vessel lumen, which in addition may be of irregular cross sectional shape (that is, not circular) but is only measured in one longitudinal plane. We agree with the authors that the analysis software is of great importance and may be a key factor in dealing with this complex situation. In spite of our reservations we, and many other workers concerned with orbital circulation, are in urgent need of a reliable solution for volumetric blood flow determination in the orbit. The results of Orge and co-workers show that we are probably not far from a definitive solution.