Efforts to analyse the complex vasculature of the eye have been frustrating to say the least. We still must rely very much on fluorescein angiography. Laser Doppler flowmetry has not yet yielded the results that had been hoped for. The same has been true of colour Doppler imaging. In a recent attempt on my part to make sense of many techniques used to study the circulation of the optic nerve, my conclusion was that much work still needs to be done.1 Harris and colleagues are to be commended for relentlessly struggling to find better ways to determine how blood flow can change under a variety of clinical settings using several techniques, especially colour Doppler imaging.
Until recently, colour Doppler imaging has been limited to visualising blood vessels, identifying direction of blood flow, and calculating blood velocity only. Pulsatility and resistive indices provide indirect evidence of resistance at or nearby the ultrasound probe, but volumetric blood flow (amount of blood/time) has not been measurable. The main problem with accurately assessing the orbital vessels has been the small size of the ophthalmic artery and especially the central retinal and ciliary arteries. In the paper by Orge et al in this issue of the BJO (p 12161216) the ophthalmic artery, because of its relatively larger size, was measured with regard to diameter in order to calculate volume and, ultimately, blood flow. However, it is evident that these investigators are still wrestling with considerable variability in measurements within their own laboratory.
The limitations of the authors’ methods are reflected in their interobserver coefficients of variability of 40% and intraobserver variability of almost 30% for calculations of blood flow. This wide range of measurements may simply have been due to the small size of the ophthalmic artery, but it may also be due to its tortuous course and the fact that it contains many branches. Among different individuals, ophthalmic arteries also have large variations in size and configuration, which will inherently make comparing blood flow between different patients subject to error. These issues require that the technicians performing each study employ a great deal of judgment, which in turn adds another layer of variability to any study, especially if this technique is to be applied to patients with disease.
With regard to size of the artery, previous efforts to measure blood flow, using similar methods employed by Ogre et al, have been made with veins, particularly shunts for renal dialysis. Even with this application, accuracy is limited. With respect to measuring arteries elsewhere in the body, there is evidence that large vessels may be accurately measured. However, the error in measuring smaller vessels is obvious. In one study, mean flow and standard errors in the common femoral artery was calculated to be 284 (SE 21) ml/min, whereas in the dorsalis pedis it was 3 (1) ml/min.2 As Orge et al point out, one is limited to measuring only a very few pixels on the screen. With regard to an artery as small as the ophthalmic artery, which is about 2 mm in diameter, one is basically counting about 10 pixels. If one considers that converting a diameter measurement to the measurement of area, the error would be squared. When we attempted to measure ophthalmic artery diameter in our own laboratory, we found that repeated measurements could easily be wrong by at least one or two pixels. We do not have the advantage of the modified software that Ogre et al refer to in their paper, but even when we employed a new technique called “B flow,” our accuracy in measuring ophthalmic artery diameters was still, in our opinion, poor.
The ophthalmic artery is also quite tortuous. In some individuals a consistently measurable segment of the artery can be identified, but in many others only a very short segment can be seen at any one time. Although the authors state “there is no need for assumptions about turbulence of flow,” “since each pixel within the image of moving blood contains a single velocity measurement,” we still remain concerned that turbulence of blood flow within the ophthalmic artery may add another source of error especially with regard to velocity measurements. Furthermore, the ophthalmic artery contains many branches, which are also known to affect blood flow velocity.
A variety of other issues have been raised by others involved in attempting to measure blood flow velocity using Doppler ultrasound. These include non-uniform insonation of the blood vessel, differential attenuation between soft tissue and blood, intrinsic spectoral broadening, frequency dependent scattering, high pass filtering designed to reduce high amplitude, low frequency, Doppler shifts due to vessel wall motion, sheer rate and haematocrit, as well as poor signal to noise ratio.2
Although Ogre et al show that blood flow within the ophthalmic artery can be estimated, variability, due to the small numbers generated from changes measured in small arteries, limits colour Doppler imaging to more qualitative rather than accurate quantitative analysis of ophthalmic artery blood flow. Even if the resolution of the method could be refined, the problem of the tortuosity and the variable course of the ophthalmic artery remains an issue. Finally, blood flow in any blood vessel is subject to systemic haemodynamics. Harris and his coworkers may have reached the point where blood flow can be measured in the ophthalmic artery to a limited degree. However, at this time, we should remain cautious regarding application of this method to clinical studies.
REFERENCES
- 1.Hedges TR. Assessing the vascular supply of the optic disc. Presented at the North American Neuro-ophthalmology Society Meeting, Rancho Mirage, California, 21 February 2001.
- 2.Holland CK, Brown JM, Scoutt LM, et al. Lower extremity volumetric arterial blood flow in normal subjects. Ultrasound Med Biol 1998;24:1079–86. [DOI] [PubMed] [Google Scholar]