Implementation of the computer tomography parallel algorithms with the incomplete set of data.

Area of Science:

• Medical Imaging
• Computational Science

Background:

• Computer tomography (CT) applications often require complete and high-quality scan data.
• Incomplete data sets, arising from difficult object access, pose significant challenges for CT reconstruction.
• Previous research indicated CT algorithms can handle incomplete data, but reconstruction time is a major limitation.

Purpose of the Study:

• To examine the practical usefulness of a proposed parallel-block approach for CT reconstruction with incomplete data.
• To address the problematic reconstruction times associated with incomplete data CT scans.
• To investigate the convergence and applicability of parallel CT algorithms in real-world scenarios.

Main Methods:

• Dividing the system of linear equations into blocks for parallel processing by different threads.
• Examining the parallel-block approach through theoretical analysis and practical implementation.
• Investigating the selection of optimal reconstruction parameters for incomplete data sets.
• Analyzing the convergence properties of the implemented parallel algorithm.

Main Results:

• The parallel-block approach is effective for CT reconstruction even with incomplete data.
• Optimal reconstruction parameters can be selected, allowing for a specified maximum error for a given number of pixels.
• The real implementation of the parallel algorithm demonstrates convergence.
• The study highlights key differences between classical and the examined incomplete data CT problems.

Conclusions:

• The parallel-block approach offers a practical and convergent solution for reducing CT reconstruction time with incomplete data.
• This method enhances the applicability of CT in situations with limited data acquisition.
• The findings support the real-world utility of parallel computing in advanced medical imaging techniques.