Among some clinical communities, ultrasound is perceived as the poor relation when it comes to medical imaging (see Diagnostic ultrasound: who pulled the plug?). Its chief advantages - safety, cost-effectiveness and versatility - have led to it being used widely by non-specialists, while being shunned by many radiologists in favour of "more glamorous" modalities like CT and MRI.

But things are now looking up again for ultrasound imaging. Thanks largely to the efforts of Smith's group at Duke, ultrasound has been reincarnated with a third dimension. Being able to produce 3D images in real-time allows clinicians to observe and measure the shape and volume of patients' internal anatomy in unprecedented detail.

Since 1987, when Smith and von Ramm patented the first high-speed 3D ultrasound system, the technology has almost completely taken over the obstetrics market in the US, with cardiology applications not far behind. Michelle Jeandron spoke to Smith to get his perspective on where this evolving technology is heading.

MJ: What are the advantages of being able to do ultrasound imaging in three dimensions?
SS: For foetal imaging, the big advantage has been in looking at facial and cranial abnormalities, and being able to measure the volumes of structures in the foetus. Also, 3D imaging allows you to measure things in directions that are not available in a normal 2D image. Cardiac 3D ultrasound is still growing, but it seems that the main advantages so far are being able to measure the volume of the left ventricle - otherwise known as the stroke volume or ejection fraction - and for guiding interventional devices, such as catheters, into the heart.

I think that 3D ultrasound will also be very valuable in places where you need real-time information - i.e. in the operating room or cardiac catheterization lab, where you don't have access to CT or MRI and you certainly don't have real-time imaging. Probably within a few years, 3D ultrasound [technology] will be small enough to fit into a purse. It's not very likely that CT will ever get that portable, and neither will MRI. So when you look at the advantages of cost, real-time and portability, ultrasound will probably always have the lead.

How does the technology actually work?
All you really need to do is move the ultrasound beam back and forth in a raster pattern, say in the x and y directions, then the depth into the tissue comprises the third dimension. If you plot the echo strength as a function of x, y and z, you have a 3D image.

What's the story behind the development of 3D ultrasound technology?
Actually, 3D ultrasound has been around since the 1950s, as a curiosity or as a research tool. For a long time, however, it was too slow to be useful for clinical applications. The speed of sound in tissue is around 1500 m/s - much less than the speed of light - so it takes a long time for the ultrasound to travel into the tissue and back up to the transducer. Then you have to move the transducer to the next spot and do it again.

Our innovation was a technology called parallel processing. This means that every time you send a pulse into the body you listen for the echoes in many different directions at once, effectively speeding up the data acquisition rate. In our case, we speeded it up by a factor of 16, meaning that we were able to make images 16 times faster than usual. As a result, we're able to create real-time 3D images. That was the birth of the current technology of high-speed 3D ultrasound.

How does the image quality compare with 2D ultrasound?
People generally accept that there is a slight degradation in the image quality due to the focusing of the transducer for parallel processing, but that's made up for by the fact that we're able to focus in all three dimensions rather than just two. For the actual task of detecting a tumour or lesion, because you're able to focus in all three directions, the loss of a little spatial resolution in two directions is compensated for by the 3D resolution. It's not clear whether there is any overall loss of quality - perhaps there is a little bit of a trade-off with high-speed 3D.

There's always room for improvement, however, and many, many people are working to improve ultrasound image quality. We've not yet reached an upper limit.

How is this technology being used commercially?
Most foetal imaging is done in 3D now - the technology of high-speed 3D ultrasound has almost totally overtaken everything in obstetrics. In cardiology also, probably a very high percentage of ultrasound scanners are 3D.

The ground swell of enthusiasm is not quite as great in cardiology as it is in obstetrics, but it's still growing. For example, a few years ago we developed the first 3D transoesophageal probe for cardiology and I think the first commercial version of that was released a few months ago. We have also made 3D catheters for cardiac applications - those have yet to be introduced commercially, although I think a number of companies are working on that. As time goes on, I think you'll see every 2D application supplanted by a 3D probe.

What is your team working on at the moment?
One of our current projects is looking at 3D ultrasound imaging of the brain - the cerebral vessels - which hopefully can be used as a diagnostic tool for stroke. Another project is to build a 3D transducer into the tip of several implantable devices. An example would be the so-called vena cava filter that filters out blood clots from the body, which is currently implanted via an endovascular approach using fluoroscopy. We think that we can actually integrate a 3D ultrasound probe into the implantation tip and hopefully get good images without exposing the patient to X-rays.

Basically, we're trying to look at every little device that's implanted into the body and see whether we can incorporate a 3D transducer into that device to make the implantation easier.

Looking ahead 20 or 50 years, how do you envisage 3D ultrasound being used?
I think it will have totally supplanted 2D ultrasound. Everywhere 2D is being used now there will be 3D, and it will be in portable devices that are as small as a laptop or a PDA.

The other area that we're working on is incorporating ultrasound into robotic surgery. The big breakthrough there would be if there was an autonomous robot that could do an ultrasound scan and then perform the surgery with the information that it had found using the 3D ultrasound. Looking ahead in a blue-sky way, that's what I see in the distant future.