The Argonne team uses artificial intelligence to decode X-ray images faster, which could foster innovation in medicine, materials, and energy.
The new computer system can reconstruct images from many X-ray data at speeds hundreds of times faster than existing ones and learn from experience, and develop more efficient ways to count and reconstruct data.
In an article published in the journal Applied Physics Letters, a US Department of Energy (DOE) computer science team at Argonne National Laboratory demonstrated the use of artificial intelligence (AI) to accelerate image reconstruction based on coherent X-ray scattering data.
The process of using computers to assemble images from coherent scattered X-ray data is called ptycography. Scientists have used a neural network that learns to transform this data into a consistent form. Hence the name of their innovation: PtychoNN.
Using artificial intelligence techniques, the research team has demonstrated that computers can be taught to predict and reconstruct images from X-ray data. They can do so 300 times faster than the traditional method.
It’s worth noting that instead of using simulated images to train the neural network, the team used real X-ray data.