Intravascular Doppler OCT

Add One

Doppler Optical Coherence Tomography (Doppler OCT) allows us to measure blood flow on a microscopic scale. Sometimes we only want to distinguish between what is moving and what isn’t, perhaps so that we can distinguish blood vessels from surrounding tissue. On other occasions we might want to actually estimate the direction and velocity of the blood flow. Either way, Doppler OCT has rarely been used in vivo with endomicroscopes, partly because of the difficulties in using phase-based techniques outside the comfort of the optics lab. A recent paper in Biomedical Optics Express (an open access journal) has reported the use of a commercial OCT endomicroscope to obtain Doppler OCT images of intra-vascular blood flow in a pig. It shows that the phase stability problem can be overcome using simple subtraction techniques, paving the way for rapid clinical translation.

The Doppler Effect

The Doppler Effect is something we’re all familiar with. As an ambulance passes by at speed, we don’t just hear a change in the volume of the siren, but also a change in its tone. Sound waves coming from objects that are moving towards us at are heard at higher pitch, and sound from objects that are moving away is heard at a lower pitch. This change in the pitch – which really means a change in the frequency of the sound wave – is approximately proportional to the velocity of the object.

Our everyday experience of the Doppler Effect is with sound, but something very similar happens with light waves. There is a shift towards the blue part of the spectrum (higher frequencies) when light is reflected off an object that is moving towards us, and towards the red part of the spectrum (lower frequencies) if the object is moving away. In fact it’s by looking at the red-shift from distant starts that astrophysicists can tell that the universe is expanding. Closer to home, it’s also the idea behind Laser Doppler Velocimetry, a technique used to measure flow speeds. Unfortunately it doesn’t work well for biological imaging because it can only measure the velocity of a single layer. This is where OCT steps in.

Depth Resolved Doppler

At first glance, Doppler OCT doesn’t seem to involve Doppler frequency shifts at all. That’s because we don’t directly measure the change in frequency, which would be tiny, but instead detect a change in phase. In figure 1(a), you can see a simulation of spectral interferogram collected by the spectrometer of a Fourier domain OCT system. This interferogram encodes a single axial or depth scan, known as an ‘A-scan’. Many of these A-scans are acquired at difference transverse positions in order to assemble a complete cross-section or ‘B-scan’.  In the example of figure 1, an object containing a single, highly reflective surface has been imaged, and the spectrum is modulated at a frequency which depends on the depth of this surface. For a more complex object containing a continuous distribution of reflecting layers then the interferogram will look much more complicated. It will be a sum of many modulations of different frequencies.

Phase shift in spectral interferogram when the reflector moves

Figure 1 : (a) A ‘spectral interferogram’ from an OCT system, encoding a single depth scan. This example shows a single reflecting surface. (b) A phase shift (red line) occurs when the object moves in between two A-scans.

If the surface moves a small distance between one A-scan and the next, there is a change in the phase of the modulation – this is shown in figure 1(b). If we can measure the change in the phase then we can calculate how far the object has moved between the two scans. Then, if we know the time between the two scans, the object’s velocity is simple to calculate. We can obtain a phase shift measurement for each frequency of modulation, corresponding to reflectors at different depths in the sample. So we can assemble a depth-dependent velocity map which, with the addition of transverse scanning, can be extended to a full 2D cross-section or even a 3D Doppler volume.

That’s the theory, but the main difficulty in actually using Doppler OCT is that the phase of the spectral interferogram is affected by things other than the flow we want to measure. Motion of the probe, bulk motion of the tissue and any instability of the phase of the light source will all introduce artefacts which can easily dominate over the useful signal. The main contribution of the paper by Sun et al. 1 was to show that these artefacts can be overcome even for catheter based Doppler OCT systems.

Intravascular Doppler

The authors used a commercially available OCT catheter – the snappily named C7-XR by St Jude Medical. This device doesn’t offer Doppler OCT, so they had to build their own signal processing system. Unlike most OCT systems this kind of probe doesn’t using galvo mirrors for its transverse scanning. Instead, A-scans are generated at a 70 degree angle to the axis of the catheter – essentially forming radial scans. The catheter is then rotated to generate a circular scan from multiple radial scans. If that’s hard to visualise, take a look at their website to see what the images look like.

Ideally we would like to collect a number of A-scans at one location, calculate the Doppler OCT, move to the next transverse position and then repeat until we have a complete Doppler velocity map. Unfortunately, it’s very difficult to build a system which can perform this kind of stop-start scanning and still maintain a usable frame rate. The rotating catheter design of the C7-XR certainly can’t do this. So the authors had to rely on using a very high scan rate relative to the rotation speed of the catheter. This means that there is large overlap between adjacent A-scans, giving a good approximation to the ideal stop-start scanning.

The authors then had to overcome the problem of artefacts resulting from the motion of the fibre optic probe as it rotates. They noticed that the artefacts were constant with radial distance, as we might expect. Luckily, OCT images from the C7-XR show the wall of the catheter as a circle towards the centre of the image. The authors started by extracting the Doppler velocity of the catheter wall for each rotational position. This told them how much of the Doppler signal was due to the fibre optic probe moving relative to the catheter. They found that if they subtracted this velocity from the velocities calculated for all radial distances then the artefacts were suppressed.

Phase Wrapping

One problem common to most implementations of Doppler OCT is known as ‘phase wrapping’. We’ve said that a change in the position of a reflector results in a phase shift proportional to that change. But we can only measure a phase shift unambiguously when the shift is less than a full cycle – less than 2π. If this shift is greater than a full cycle – say a cycle and a quarter (5π/4) – then there is no way to tell if the shift has been a quarter of a cycle (π/4), a cycle and a quarter (5π/4), two cycles and quarter (9π/4) and so on. Usually we just assume that the shift was less than a whole cycle (2π), which basically sets an upper limit on the velocity we can measure.

There’s a way round this problem for certain situations, and intravascular blood flow happens to be one of them. We start by making an assumption that the blood is flowing in a physically reasonable way. We expect the blood to flow fastest in the centre of the vessel, and slowest near the walls. And there should be a well-behaved velocity gradient in-between, by which we mean that there won’t be one little spot where the blood is for some reason flowing ten times faster than everywhere else!

Armed with this assumption we can do something called ‘phase-unwrapping’. Take, for example, a point where the flow is sufficiently fast so that the phase shift is just less than one cycle. If a neighbouring point has a phase shift which appears to be just greater than zero, we can assume that, rather than the velocity having suddenly dropped to zero, that the phase has wrapped. So the the true phase shift was one cycle plus a little bit, and we use this larger phase shift to calculate the velocity. This is, at least, the generally philosophy behind phase unwrapping – in practice, of course, it’s all done mathematically. Using this idea, Sun et al. were able to measure velocities of up to 50 cm/s.

Problems Remain

I think this paper shows quite nicely that Doppler OCT can be used to measure intravascular blood flow. Having said that, the authors do acknowledge that a couple of problems remain. Firstly, Doppler OCT can only measure motion parallel to the optical axis (i.e. in the direction of the A-scan). For this type of OCT catheter the blood flow is almost entirely perpendicular to the A-scan direction. Luckily the OCT catheter was designed to have an A-scan direction which is 70 degrees, rather than 90 degrees, to the catheter axis. The catheter will also tends to lie at a slight angle with respect to the axis of the artery. But if we don’t know exactly what this angle was then it’s impossible to work out the true velocity of the blood flow.

This angle problem could potentially be overcome by finding some way to measure the orientation of the catheter – in fact the authors showed that this can be done in principle by first acquiring a 3D OCT scan using a pullback. But another, more fundamental, issue remains. The presence of the catheter in the artery is obviously going to drastically affect blood flow, meaning that the measurement takes places under highly artificial circumstances. It may be the measurement is still useful, but it certainly makes the technique less attractive.


  1. Sun, C., et al., In vivo feasibility of endovascular Doppler optical coherence tomography. Biomedical optics express, 2012. 3: p. 2600-10.

Leave a Comment

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>