Vortrag: "Two-stage approach for low-dose and sparse-angle CT image reconstruction using backprojection"

Der Abstract fasst die Inhalte zusammen:

This talk presents a novel two-stage approach for computed tomography (CT) reconstruction, designed to minimize radiation exposure in sparse-angle and low-dose settings while preserving high image quality. The two-stage approach consists of an initial reconstruction followed by a neural network for image refinement. In the initial reconstruction, we apply the backprojection (BP) instead of the traditional filtered backprojection (FBP). This choice improves computational speed and offers advantages in handling more complex geometries, such as fan-beam and cone-beam CT. Additionally, BP addresses noise and artifacts in sparse-angle CT by leveraging its inherent noise-smoothing effect, which reduces streaking artifacts common in FBP reconstructions. In the second stage, we enhance the reconstruction using DRUNet, a pretrained Gaussian denoiser by Zhang et al., fine-tuned for this task. We call this method BP-DRUNet and evaluate its performance on both a synthetically generated ellipsoid dataset and the well-established LoDoPaB-CT dataset. Our results demonstrate that BP-DRUNet achieves competitive PSNR and SSIM scores compared to FBP-based methods, delivering visually comparable results across all tested angular setups.