The correct image

Edinburgh University researchers have adapted an algorithm developed to analyse spectrum data from distant galaxies to help correct MRI images if a patient moves while a scan is in progress.

Prof Alan Heavens of Edinburgh’s theoretical astrophysics department said: ‘If a volumetric image of the head is taken, a subsequent one may be out of line because the patient has moved. We can take a sequence of quick exposures, line them up and add them together, which improves image quality. If a patient moves his or her head, we know exactly how much it is by, and correct for that in the images.’

Research fellow Ben Panter added: ‘We use the whole volume of the data and rather than look for individual features in the data we compress it down, applying the algorithm to choose the right compression. The earlier application was analysing one-dimensional data, but in this new application we’re addressing volume 3D data.’

The time to align the images is less than the time it takes to acquire them, so it effectively takes place in real-time. The researchers are currently talking to medical technology companies with a view to incorporating the software into scanners.

The study is funded by a Scottish Enterprise proof of concept grant, and last week it won a Research Councils UK business plan competition for best commercialisation proposal.

Future applications could include facial recognition, hyperspectral imaging or quality control, where an image of an object is compared to a control sample.