F-GMT11-1: Real time ultrasound compounding

Ultrasound images (B-scans) represent the amount of sound reflected from a cross-section of anatomy. However, sound is attenuated as well as reflected as it passes through the anatomy, and also undergoes multiple additional reflections at material interfaces. Both of these effects make the B-scan very hard to interpret.

However, artefacts are in general dependent on the direction of insonification (the direction of the ultrasound 'beam'), whereas real data is not. This direction is controlled by a 'phased array' in which several of the ultrasound transducer elements are fired at slightly different times, so that the combined beam focuses at a particular location.

By modifying the delays at each element, it is possible to stear the ultrasound beam so that it effectively fires at an angle to the transducer, rather than in an orthogonal direction. If multiple beams are fired sequentially at different angles, the effect will be as if the same anatomy were scanned from multiple directions.

This project will investigate the extent to which artefacts in B-scans can be detected and displayed by using such multi-angle data. The first part of the project will investigate the possible range of angles to which the ultrasound beam can be steered, the second will investigate how such data can best be combined. The aim is to generate a visualisation of each B-scan in which the likelihood of each part of the image being real or artefact is visualised in some way, perhaps by using a colour overlay.

The project will involve the use of diagnostic ultrasound machines and draw on techniques in signal processing. It will require experience of Matlab and/or C/C++ programming.

Ultrasound beam steering or 'forming' is achieved by firing neighbouring elements with different delays. In commercial ultrasound machines, this is nearly always done symmetrically, as above. This project will investigate asymmetrical beam steering.

An ultrasound image, like the one above, does not reveal any direct material properties: the intensity of the image at any given point depends on the nature of the tissue at that point, and also on a variety of phenomena involving the surrounding tissue. This project will attempt to improve this situation by using multiple views to clarify which parts of the above image are artefact.