NIST is developing a novel neutron interferometric phase imaging method using a grating-based, far-field interferometer with 1000× increase in time resolution and 10× improvement in spatial resolution over prototypes. Such an instrument enables measuring the structure of heterogeneous systems including aging concrete, blood clots, highly-efficient batteries, multiphase flow in colloidal and geological systems, and the burgeoning field of metal 3D printing. As this far-field neutron interferometer will be generating terabytes of image data per day, there are several open research problems that would be the research focus of a postdoctoral candidate. The problems include finding optimal imaging configurations, accelerated 3D tomographic reconstruction from multiple 2D images and multiple channels, optimization of 2D projection viewpoints (dose reduction and time savings), the use of artificial intelligence and traditional machine learning models for 3D volume segmentation, designing tools for 3D segmentation annotation and verification, registration of 3D volumes, evaluations of accuracy and uncertainties of image analyses, visualization of multiple terabyte-sized 3D volumes, quantification and statistical analyses of volume characteristics, and dissemination of petabyte-sized 3D image collections with web-based visual inspection and browsing capabilities. A small subset of project relevant publications is listed below. A candidate should have at least a master’s degree in computer science or related field and have a background in imaging, image analyses, machine learning, and software engineering.
(1) A. J. Brooks et al, “Neutron interferometry detection of early crack formation caused by bending fatigue in additively manufactured SS316 dogbones,” Materials and Design, 140 (2018) 420–430 https://doi.org/10.1016/j.matdes.2017.12.001
(2) T. Pipatsrisawat et al, “Performance Analysis of The Filtered Backprojection Image Reconstruction Algorithms,” ICASSP 2015.
(3) D.M. Pelt and J.A. Sethian, “A mixed-scale dense convolutional neural network for image analysis,” PNAS, January 8, 2019.