首页|New approach to enhance multi-material computed tomography reconstructions by selecting the optimal combination of scanning orientations for multiple scan fusion
New approach to enhance multi-material computed tomography reconstructions by selecting the optimal combination of scanning orientations for multiple scan fusion
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
Elsevier
The use of multi-positional X-ray Computed Tomography (XCT) has been shown to reduce metal artifacts when scanning multi-material assemblies. However, its effectiveness heavily relies on the choice of scan orientations to fuse. This paper presents a novel approach to determine the optimal combination of scan positions by taking a small number of simulated projections and deriving projection-based metrics to estimate the quality of the resulting data fusion. Two regression models, linear and Random Forest, were used and compared for prediction of the optimal combination. Results on an adaptation from an industrial case study showed that, while both models provide predictions that are well correlated to the actual ranked values on the test data (Spearman's rho of 0.841 for the linear model and 0.964 for the Random Forest), the Random Forest model performed better than the linear model on all the proposed evaluation metrics, and consistently ranked high quality combinations among its top predictions. These results highlight the effectiveness of the approach in finding optimal combinations of scan positions with little input data, providing a time-efficient solution for improving XCT reconstruction of multi-material objects.
NETL
NSTL
Elsevier
