METAL ARTIFACT CORRECTION OF CONE BEAM CT IMAGES OF CORRUGATED TUBES BASED ON A GENERATIVE ADVERSARIAL NETWORK

JIANG Shibo,SUN Yuewen,Zhang Jiali,Zhang Huaxia,Zhang Zehuan,Wu Zhifang

DOI: https://doi.org/10.1299/jsmeicone.2023.30.1280

2023-01-01

The Proceedings of the International Conference on Nuclear Engineering (ICONE)

Abstract:Intermediate heat exchangers transfer high-temperature heat from the very-high-temperature gas-cooled reactor (VHTR) to the process heat application facility such as hydrogen production, which is a key component in nuclear heat utilization. To ensure that the IHX provides sufficient heat for process heat applications, different types of corrugated tubes are used to enhance the heat transfer capacity of IHX. Precise geometry parameters are crucial in the study of the thermalhydraulic performance of the corrugated tubes. Using the highprecision industrial cone beam CT can achieve the nondestructive and high-precision measurement, and obtain the precise geometric parameters of the tubes inner surface. However, due to the strong attenuation characteristics of metal materials on X-ray, there are serious scattering, hardening and statistical noise in CT when scanning the corrugated tubes. These factors lead to strong metal artifacts in the reconstructed images, which seriously affect subsequent process such as segmentation and parameter measurements. We propose a method based on generative adversarial networks in order to correct metal artifact while preserving the details of the CT image features well. The proposed method generates CT image samples through simulation methods and obtain the mapping relationship between metal-artifact-containing and metalartifact-free images by training. Validation of simulated CT images and corrugated CT image datasets, the method proposed in this paper preserves the feature information such as CT image size and boundary while correcting metal artifacts well. And it provides accurate parameters for further research on the thermal-hydraulic performance of corrugated tubes, reduces the differences or deviations between experiments and numerical simulations..