Element-specific determination of X-ray transmission signatures using neural networks

C.R. Day, J.C. Austin, J.B. Butcher, P.W. Haycock, A.T. Kearon

Research output: Contribution to journalArticle

Abstract

In this article, we report on the application of neural networks to the problem of making an element-specific determination of unknown metal targets based on the characteristics of their transmitted X-ray signatures. Our method was applied to two groups of metal targets that we characterised as light elements (atomic numbers between 40 and 50) and heavy elements (atomic numbers between 73 and 83). In all cases their X-ray signatures were pre-processed; randomly allocated into training and testing datasets; and then presented to a self-organising map neural network in order to perform the element-specific determination. The technique was able to make a correct determination of unknown metal targets with an accuracy of 95% for the heavy elements and 99% for the light elements.
Original languageEnglish
Pages (from-to)446-451
JournalNDT & E International
Volume42
Issue number5
DOIs
Publication statusPublished - Jul 2009

Fingerprint Dive into the research topics of 'Element-specific determination of X-ray transmission signatures using neural networks'. Together they form a unique fingerprint.

  • Cite this