It was almost 52 years ago that engineer Godfrey Hounsfield and radiologist James Ambrose carried out the first computed tomography (CT) scans, extending X-ray technology beyond planar 2D images into the third dimension. This was a revolutionary development in medical imaging which in 1979 saw Hounsfield and physicist Allan Cormack share the Nobel Prize in Medicine. In the field of ultrasound there have also been key advances in the past half century, with 3D images of a foetus in its mother’s womb impressing both clinicians and parents alike. 3D ultrasound – so, how’s that work?

With a traditional linear ultrasound transducer, 2D images are acquired where the plane is determined by the angular orientation of the probe. In freehand mode, the sonographer can tilt the probe, thus capturing a series of 2D ultrasound “slices” which may be reconstructed into a 3D dataset if the angle of the probe is known. The tilting of the acquisition plane can be automated with a small motor located in the probe (seen at left), and this in fact is how most of the 3D foetal images (seen above right) are produced.

Instead of tilting the transducer, it is also possible to translate the probe in a linear fashion, thus creating a series of parallel slices where the distance between slices in known. This is the basis for the 3D automated breast ultrasound (ABUS) systems marketed by GE Healthcare and Siemens, and implemented in CapeRay’s Aceso system which combines 3D ultrasound with 2D mammography (see image at right, where the ultrasound probe is shown in blue).

While most ultrasound systems are based on sound waves being reflected, it is also possible to acquire 3D images when the waves are transmitted through the breast. This is the approach taken by Delphinus and QT Ultrasound in which the woman lies face-down on a bed, and there is a transmitter that generates ultrasound waves which pass through the breast into a receiver (see diagram at left, © AAPM).

While the aforementioned methods to generate 3D ultrasound images are all based on transducers that are essentially one-dimensional and rely on movement of the probe, there is another approach in which the ultrasound elements are stationary and arranged in a 2D matrix (see image at right). Although this does provide significant technical challenges – since there are now N² elements for the system to monitor – it is possible to mitigate these problems by employing either phased arrays or even a row-column addressed strategy. To be sure, 3D ultrasound has some way to go before it can approach the breakthrough of CT.