Deep Learning in Science and Engineering
- Texas A&M University
- College Station, TX
- Joe C. Richardson Petroleum Engineering Building (RICH) 910
- Peter Kuchment, Department of Mathematics
- Deep Learning in Detecting Illicit Nuclear Materials
Abstract
The development of methods for source detection in high noise environments is an important topic in single-photon emission computed tomography (SPECT) medical imaging and homeland security applications. The detection of low emission nuclear sources in the presence of significant background noise (SNR > 0.01) is of great interest since such a robust detection system can prevent the smuggling of weapons-grade nuclear material. A source detection method based on the analysis of data obtained from Compton type cameras and their analogs using deep learning is developed and evaluated, and compared to previous statistical detection techniques.