Abstract Characterizing porous metals requires reliable 3D reconstruction. We compare X-ray computed tomography (XCT) and serial sectioning using a virtual ground truth from segmented XCT data. A GPU-based virtual XCT pipeline simulates imaging artifacts, serial sectioning is modeled via extracted topographies and simulated annealing reconstruction. XCT achieves over 98% accuracy but remains sensitive to segmentation thresholding. Serial sectioning captures visible surfaces precisely but cannot measure occluded regions, yielding lower reconstruction accuracy of 91%. Given these complementary strengths, we propose a hybrid approach where CLSM-derived porosity measurements inform XCT threshold selection, grounding this critical parameter in measured sample properties rather than image statistics alone.
Friedrich et al. (Mon,) studied this question.