Efficient and effective regularised ART for computed tomography.

Area of Science:

• Medical Imaging
• Computational Science

Background:

• Algebraic Reconstruction Technique (ART) is a computed tomography method with advantages in handling incomplete projection data.
• Regularized ART (RegART) reduces noise but suffers from high computational complexity.
• Filtered Back Projection (FBP) is a common alternative in computed tomography.

Purpose of the Study:

• To develop a computationally efficient version of RegART.
• To leverage nVidia's CUDA technology for accelerating RegART.
• To compare the performance of the fast RegART against FBP.

Main Methods:

• Implementation of a fast Algebraic Reconstruction Technique (ART) using nVidia's CUDA.
• Regularization techniques were applied to enhance image quality.
• Comparative analysis with Filtered Back Projection (FBP) was performed.

Main Results:

• The CUDA-accelerated RegART demonstrates significantly reduced computational complexity.
• The developed method achieves favorable performance compared to FBP.
• Noise levels are substantially reduced, consistent with RegART's benefits.

Conclusions:

• Fast RegART using CUDA offers an efficient and effective solution for computed tomography reconstruction.
• This approach overcomes the computational limitations of traditional ART and RegART.
• The method shows promise for improved medical imaging applications.