DLBayesian: An Alternative Bayesian Reconstruction of Limited-view CT by Optimizing Deep Learning Parameters

TL;DR

DLBayesian is a novel deep learning-based Bayesian reconstruction method for limited-view CT that enhances artifact removal and generalization by integrating data-driven priors with data consistency constraints, validated on real and simulated data.

Contribution

It introduces a three-stage DL-driven Bayesian reconstruction framework that optimizes network parameters, evaluates their significance, and adaptively refines the reconstruction for diverse cases.

Findings

Outperforms existing DL methods in artifact removal and generalization.

Demonstrates superior results on sparse-view and missing data CT cases.

Validates effectiveness in practical experiments on a dead rat.

Abstract

Limited-view computed tomography (CT) presents significant potential for reducing radiation exposure and expediting the scanning process. While deep learning (DL) methods have exhibited promising results in mitigating streaking artifacts caused by a reduced number of projection views, their generalization remains challenging. In this work, we proposed a DL-driven alternative Bayesian reconstruction method (DLBayesian) that efficiently integrates data-driven priors and data consistency constraints. DLBayesian comprises three stages: group-level embedding, significance evaluation, and individual-level consistency adaptation. Firstly, DL network parameters are optimized to learn how to eliminate the general limited-view artifacts on a large-scale paired dataset. Then, we introduced a significance score to quantitatively evaluate the contribution of parameters in DL models as a guide for the subsequent individual-level adaptation. Finally, in the Bayesian adaptation stage, an alternative Bayesian reconstruction further optimizes the DL network parameters precisely according to the projection data of the target case. We validated DLBayesian with sparse-view (90 views) projections from a circular trajectory CT and a special data missing case from a multi-segment linear trajectory CT. The results underscore DLBayesian's superior generalization capabilities across variations in patients, anatomic structures, and data distribution, as well as excelling in contextual structure recovery compared to networks solely trained via supervised loss. Real experiments on a dead rat demonstrate its capability in practical CT scans.