Computed tomography medical image reconstruction on affordable equipment by using Out-Of-Core techniques

Area of Science:

• Medical Imaging
• Computational Science
• Image Reconstruction

Background:

• Computed Tomography (CT) scans are vital medical tools.
• Reconstructing high-quality CT images with reduced radiation is a key challenge.
• Existing algebraic factorization methods for CT reconstruction are computationally intensive, limiting clinical application.

Purpose of the Study:

• To develop a faster and efficient method for CT image reconstruction.
• To address the high computational cost associated with algebraic factorization techniques.
• To enable high-quality CT image reconstruction suitable for daily clinical practice.

Main Methods:

• Proposed a novel CT image reconstruction method utilizing QR factorization.
• Implemented Out-Of-Core techniques for efficient processing on standard multicore processors and Solid-State Drives (SSDs).
• Compared the QR method against Least Squares QR (LSQR) and Filtered Back Projection (FBP) algorithms.

Main Results:

• Achieved rapid and high-quality CT image reconstruction using the proposed QR method on affordable hardware.
• Evaluated image quality using Peak Signal-To-Noise Ratio (PSNR) and Structural Similarity Index (SSIM), yielding high scores.
• Demonstrated superior Input/Output performance with SSDs compared to traditional spinning disks.

Conclusions:

• The developed QR factorization method and software efficiently solve large-scale systems for CT reconstruction.
• The approach enables high-quality CT image reconstruction with significantly reduced computation time.
• This method offers a viable solution for accelerating CT image reconstruction in clinical settings.