Community Research and Development Information Service - CORDIS

FP7

F.A.U.S.T. Result In Brief

Project reference: 325661
Funded under: FP7-PEOPLE
Country: United Kingdom

New tool reduces uncertainties in medical imaging

Subject motion is a pervasive problem in magnetic resonance imaging (MRI), especially when attempting to assess foetal well-being. An EU-funded research team used advanced software that assesses image accuracy and presents more information, including errors and uncertainty, while also allowing fast image reconstruction.
New tool reduces uncertainties in medical imaging
Even with modern machines, evaluating foetal MRI can be very time consuming and labour intensive. Motion-induced artefacts pose difficulties in image processing, resulting in overlaps between spatially adjacent slices of a target volume.

With few exceptions, research on medical imaging software has so far neglected visual representation of errors and uncertainties. Within F.A.U.S.T. (Flexible application of uncertainty for scanning and tracking), scientists introduced a hardware-accelerated framework that enables measurement and real-time visualisation of uncertainties while also speeding up computational time. This new software tool should allow earlier detection of congenital abnormalities as well as abnormal development following preterm birth.

Scientists presented slice-to-volume reconstruction methods for obtaining high-quality 3D images using graphics processing unit (GPU) hardware acceleration. A key feature of the newly developed framework is the use of flexible point spread functions that allow high reconstruction accuracy from motion-corrupted image stacks. It also integrates approaches to estimating the amount of motion in corrupted image stacks and automatically selecting the stack with the least amount of motion. This approach eliminates the need to manually prepare the data.

The software was designed with a strong focus on GPU scalability, leading to computation times significantly faster than those possible with available methods. It thus helps overcome long computational times in medical image analysis, especially those arising from imaging moving foetuses in utero with MRI.

Accurate interpretation of foetal MRI can provide valuable information that helps prenatal counselling, guides therapy and supports research studies. Project activities and results were published in numerous peer-reviewed journals.

Related information

Keywords

Medical imaging, MRI, foetal, image reconstruction, hardware acceleration
Record Number: 175277 / Last updated on: 2016-03-08
Domain: Biology, Medicine