Non-linear dual-energy method development and evaluation for industrial computed tomography

Abstract

Industrial computed tomography of multi-material objects can be problematic due to difficulties in optimising the x-ray spectra of the scan. A possible solution to the problem is to use two x-ray spectra when scanning objects. The results from such scans can be fused into a single data-set that contains enhanced information. This practice is known as dual-energy computed tomography (DECT). In this work, the aim was to investigate two DECT methods ability to improve measurements in multi-material phantoms. To determine the performance of the methods three different phantoms containing precision spheres as measurement objects were investigated. To improve measurements in this work was defined as improving the measurement consistency of diameter measurements. The phantoms were also scanned with a single setting for comparison. The fusion of the data-sets was done using two methods, a linear fusion, and a novel non-linear fusion. Both of the methods relies on pre-reconstruction fusion of data-sets. The results show that both of the DECT methods improved measurement results significantly compared to the reference method. Further, the results show that the novel non-linear DECT method produces more accurate measurement results compared to the linear method.