The influence of reconstruction methods on measurements in CT-volumes

Herold, F.; YXLON International GmbH; Germany

Kratz, B.; YXLON International GmbH; Germany
Herold, F.; YXLON International GmbH; Germany

ID: ECNDT-0157-2018
Download: PDF
Session: CT-Applications 1
Room: G3
Date: 2018-06-13
Time: 09:20 - 09:40

In industrial Computed Tomography (CT) the state-of-the-art reconstruction method for cone-beam CT is the well-known algorithm developed by Feldkamp, Davis, and Kress (FDK). Beside this standard several alternative strategies exist which involve certain advantages depending on the application of interest, e.g. the usage of prior knowledge or arbitrary trajectories by using algebraic reconstruction techniques (ART). These approaches have a higher computational complexity compared to the FDK. However, with the increasing computational power in the last years, these reconstruction techniques become more and more applicable.
Regarding metrology there are two main influencing factors for accurate measurement results: the projective geometry and the imaging chain. The quality of the projective geometry can be measured by SD and the image quality for example by the modulation-transfer-function (MTF) and contrast-discrimination-function (CDF) following ASTM E 1695. The last part of the imaging chain is the volume reconstruction method. Especially for bidirectional measurement the internal or external boundary surface has to be determined very accurate. Potential X-ray imaging artifacts due to certain material and object shape influences will lead to distortion of the found boundary surface. If a new algorithm is used these distortions might change and it is important to know the expected effect on the final measurements.
In this extended abstract the results of an ART reconstruction are compared with the FDK-results regarding the quality of the projective geometry and the imaging chain as well as varying the number of projection images. Based on this comparison a conclusion is possible how metrological applications may be influenced by a chance of the reconstruction strategy and a reduction of the number of images.