With scientific experiments producing increasing amounts of data, the demand for significant processing power is higher than ever and can only be achieved through data intensive techniques.
A new system called ULTRA which has been developed by computing experts at the Science and Technology Facilities Council means that within tens of minutes, scientists can have hundreds of digital replicas of a sample of interest, inside and out, depicting layers and segments of the internal details at a resolution not achievable before.
A typical 3D reconstruction which is vital for gaining a more thorough understanding of an object of interest can already be done in minutes using one of the most common and fast existing algorithms, the Filtered Back Projection (FBP) algorithm. However, where there is a need to scan an object from different angles across different energy ranges to reveal contrasts within the article, iterative algorithms are needed.
Iterative algorithms typically run slower than the FBP algorithm, taking 100s of minutes to run. However, the new ULTRA platform speeds up that process:
“Our system not only speeds up the process of reconstruction, saving hours of experimental time, it also provides a wider range of computational intensive algorithms to reconstruct scans; for example in cases where there are missing angles", said Erica Yang, Head of Visualisation at STFC's Scientific Computing Department.
The insight obtained using ULTRA can then be used to steer the next experiment step, for example, to adjust sample positions and beam alignment, or to decide whether to use different reconstruction algorithms or parameters, or image filters.
ULTRA is currently being tested with STFC's ISIS Neutron and Muon Source - a research facility housing a suite of giant microscopes, or instruments - that enable scientists to see things at the atomic level. However, it has the potential to be exploited in other experimental facilities.
Marion O'Sullivan Impact Manager, Scientific Computing Department
Erica Yang Head of Visualisation, Scientific Computing Department