A methodology for finding the optimal iteration number of the SIRT algorithm for quantitative Electron Tomography

Area of Science:

• Electron Tomography
• Image Processing
• Materials Science

Background:

• Simultaneous Iterative Reconstruction Technique (SIRT) is crucial for reconstructing sample volumes from noisy electron tomography images.
• SIRT algorithm's output accuracy is sensitive to implementation and iteration count, impacting segmentation results.

Purpose of the Study:

• To propose a methodology for selecting the optimal number of iterations for SIRT reconstruction.
• To improve the accuracy of object segmentation in reconstructed volumes.

Main Methods:

• Developed a methodology based on statistical analysis of intensity profiles at object edges.
• Utilized a phantom resembling a carbon black aggregate for validation.
• Compared SIRT implementations from TOMOJ and TOMO3D free software packages.

Main Results:

• The proposed methodology allows for the selection of an iteration number that yields the most accurate segmentation.
• Validation using a carbon black aggregate phantom demonstrated the effectiveness of the approach.
• The study assessed the impact of different SIRT implementations on reconstruction accuracy.

Conclusions:

• The developed methodology provides a reliable way to optimize SIRT reconstructions for accurate segmentation in electron tomography.
• This approach enhances the utility of free software packages like TOMOJ and TOMO3D for quantitative analysis.