A Computationally Efficient Reconstruction Algorithm for Circular Cone-Beam Computed Tomography Using Shallow Neural...

Area of Science:

• Medical Imaging
• Materials Science
• Industrial Quality Control

Background:

• Circular cone-beam (CCB) Computed Tomography (CT) is crucial for quality control and imaging.
• High-throughput CT requires fast reconstruction algorithms to balance speed and quality.
• Limited data scenarios pose challenges for traditional CT reconstruction.

Purpose of the Study:

• Introduce the Neural Network Feldkamp-Davis-Kress (NN-FDK) algorithm.
• Improve CCB CT reconstruction accuracy and computational efficiency.
• Ensure low training data requirements and fast training for high-throughput applications.

Main Methods:

• Developed the NN-FDK algorithm, integrating machine learning into the FDK algorithm.
• Compared NN-FDK against standard CT reconstruction and deep neural network artifact removal methods.
• Evaluated performance in scenarios with high noise, limited projection angles, and large cone angles.

Main Results:

• NN-FDK achieves reconstructions substantially faster than alternatives, except for standard FDK.
• The algorithm produces accurate CCB CT reconstructions under challenging conditions (noise, limited data, large cone angles).
• NN-FDK training time is significantly lower than deep neural networks, with minimal accuracy compromise.

Conclusions:

• NN-FDK offers a viable solution for improving image quality in high-throughput CT scanning.
• The algorithm maintains computational efficiency while enhancing reconstruction accuracy.
• Its low training demands make it suitable for practical, resource-constrained CT settings.