Performance verification of the "ghost"-free tomographic reconstruction method QURT

Area of Science:

• Electron Microscopy
• Materials Science
• Image Reconstruction

Background:

• Conventional tomographic reconstruction methods suffer from artifacts like ghost images due to the missing-wedge effect and require numerous projection images.
• Advanced methods such as compressed sensing often necessitate parameter adjustments or prior knowledge, limiting their applicability.

Purpose of the Study:

• To quantitatively evaluate the practical performance and resolution of the novel quantisation unit reconstruction technique (QURT).
• To demonstrate QURT's effectiveness in reducing artifacts, particularly ghost images, in various tomographic applications.

Main Methods:

• QURT reconstructs tomograms by discretely arranging grey-level quantisation units in 3D space.
• Performance was assessed using Fourier ring correlation and other metrics in simulations with test patterns.
• Experimental validation was performed using HAADF-STEM and EDXS tomography on a catalyst sample.

Main Results:

• QURT generated minimal ghost images outside the test pattern region in simulations, significantly suppressing artifacts even with the missing-wedge.
• Experimental application on a Pd/CeO₂-ZrO₂-Al₂O₃ catalyst sample clearly revealed pore structure without ghosting inside pores or loss of detail in low-contrast Al regions.
• A 3D elemental map was successfully reconstructed using EDXS tomography with only 22 tilt series, despite missing-wedge data.

Conclusions:

• QURT is a robust tomographic reconstruction method that overcomes limitations of conventional techniques by minimizing ghost images and reducing data requirements.
• The technique offers high practical performance and resolution, validated in both simulations and experimental applications like HAADF-STEM and EDXS tomography.
• QURT's ability to function without parameter tuning or prior knowledge makes it a versatile tool for advanced materials analysis.