Introduction of total variation regularization into filtered backprojection algorithm

TL;DR

This paper introduces a novel total variation regularization approach to enhance filtered backprojection (FBP) image reconstruction, demonstrating improved accuracy over traditional methods through validation with radioactive tracer images.

Contribution

The paper presents the integration of total variation regularization into FBP, offering a new method that outperforms standard regularization techniques in image reconstruction quality.

Findings

Higher cross-correlation with real images

Improved image quality over standard FBP

Validated with Derenzo phantom

Abstract

In this paper we extend the state-of-the-art filtered backprojection (FBP) method with application of the concept of Total Variation regularization. We compare the performance of the new algorithm with the most common form of regularizing in the FBP image reconstruction via apodizing functions. The methods are validated in terms of cross-correlation coefficient between reconstructed and real image of radioactive tracer distribution using standard Derenzo-type phantom. We demonstrate that the proposed approach results in higher cross-correlation values with respect to the standard FBP method.