Researchers at University College London Developed a Deep Learning-based Method to X-Ray Luggage to Detect Explosives

The development of phase-based techniques has accelerated the pace of X-ray imaging. Dark-field images are sensitive to inhomogeneities on a length scale below the system's spatial resolution, and phase contrast images are improved for detailed visibility. A new technique for X-raying luggage to find trace levels of explosives was developed by a team of researchers from University College London, Nylers Ltd., and XPCI Technology Ltd. They demonstrated how dark-field produces a texture specific to the substance being photographed and how combining it with traditional attenuation improves the ability to distinguish amongst threat materials. They have also published their work in Nature Communications journal, which involves adapting a conventional X-ray detector and using a deep-learning application to better detect hazardous chemicals in luggage.