A Look-Up Table-Based Ray Integration Framework for 2-D/3-D Forward and Back Projection in X-Ray CT

Area of Science:

• Medical Imaging
• Computational Science
• Image Reconstruction

Background:

• Iterative algorithms enhance computed tomography (CT) image reconstruction, particularly for low-dose acquisitions.
• These methods effectively mitigate image artifacts but are computationally intensive.
• Accurate CT system modeling during projection and back projection is crucial for image quality.

Purpose of the Study:

• To develop a computationally efficient framework for accurate projection of image basis functions in CT reconstruction.
• To reduce the computational complexity of iterative CT reconstruction methods.
• To improve the performance of CT image reconstruction using GPU acceleration.

Main Methods:

• A novel framework replacing complex voxel projection with memory operations.
• Pre-computed coefficients stored in look-up tables, parameterized by CT system geometry.
• GPU acceleration for efficient look-up table operations.

Main Results:

• The proposed framework significantly reduces computational cost compared to existing methods.
• Look-up tables require minimal storage (kilobytes).
• Demonstrated effectiveness in both numerical and clinical experiments.

Conclusions:

• The developed framework offers a computationally efficient alternative for iterative CT image reconstruction.
• It enables faster processing without compromising image quality, especially in low-dose scenarios.
• This approach shows promise for improving the clinical applicability of advanced CT reconstruction techniques.