eduzhai > Applied Sciences > Engineering >

BraggNN Fast X-ray Bragg Peak Analysis Using Deep Learning

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 15 pages

Abstract: X-ray diffraction based microscopy techniques such as high energy diffractionmicroscopy rely on knowledge of position of diffraction peaks with highresolution. These positions are typically computed by fitting the observedintensities in detector data to a theoretical peak shape such as pseudo-Voigt.As experiments become more complex and detector technologies evolve, thecomputational cost of such peak shape fitting becomes the biggest hurdle to therapid analysis required for real-time feedback for experiments. To this end,this paper proposes BraggNN, a machine learning-based method that can localizeBragg peak much more rapidly than conventional pseudo-Voigt peak fitting. Whenapplied to our test dataset, BraggNN gives errors of less than 0.29 and 0.57voxels, relative to conventional method, for 75 and 95 of the peaks,respectively. When applied to a real experiment dataset, a 3D reconstructionusing peak positions located by BraggNN yields an average grain positiondifference of 17 micrometer and size difference of 1.3 micrometer as comparedto the results obtained when the reconstruction used peaks from conventional 2Dpseudo-Voigt fitting. Recent advances in deep learning method implementationsand special-purpose model inference accelerators allow BraggNN to deliverenormous performance improvements relative to the conventional method, running,for example, more than 200 times faster than a conventional method when using aGPU card with out-of-the-box software.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×