Optimizing X-Ray Computed Tomography Settings for Dimensional Metrology Using 2D Image Analysis

The current way to choose X-ray computed tomography (XCT) scanning settings is usually manual and prone to operator errors. This paper presents an effective semiautomatic protocol that proves a high correlation between the local contrast-to-noise (CNR) of XCT two-dimensional (2D) projection image (prior to reconstruction) quality and the resulting XCT 3D volume scan quality. This high correlation allowed the comparison of four XCT settings to determine the one with the smallest error, solely by locally using the CNR equation on one 2D projection (prior to reconstruction) of an additive manufactured lattice structure. Verification of the protocol was done by using a workpiece and comparing the chosen XCT setting reconstructed workpiece dimensions to the ones measured using a coordinate-measuring machine (CMM). This new method can reduce the operator error and time needed to compare different XCT setting combinations. The proposed protocol is a step closer to an automated XCT parameter selection procedure, limiting user dependency and error while increasing accuracy and fidelity.