On the Benefit of Dual-domain Denoising in a Self-supervised Low-dose CT Setting
This work addresses image quality restoration in low-dose CT scans, which is crucial for medical imaging applications, by introducing a novel dual-domain approach that is incremental but offers specific gains.
The paper tackles the problem of low-dose CT image quality by proposing an end-to-end trainable reconstruction pipeline with denoising in both projection and image domains, achieving improvements of 82.4-94.1% in PSNR and 12.5-41.7% in SSIM on abdomen CT data relative to baseline.
Computed tomography (CT) is routinely used for three-dimensional non-invasive imaging. Numerous data-driven image denoising algorithms were proposed to restore image quality in low-dose acquisitions. However, considerably less research investigates methods already intervening in the raw detector data due to limited access to suitable projection data or correct reconstruction algorithms. In this work, we present an end-to-end trainable CT reconstruction pipeline that contains denoising operators in both the projection and the image domain and that are optimized simultaneously without requiring ground-truth high-dose CT data. Our experiments demonstrate that including an additional projection denoising operator improved the overall denoising performance by 82.4-94.1%/12.5-41.7% (PSNR/SSIM) on abdomen CT and 1.5-2.9%/0.4-0.5% (PSNR/SSIM) on XRM data relative to the low-dose baseline. We make our entire helical CT reconstruction framework publicly available that contains a raw projection rebinning step to render helical projection data suitable for differentiable fan-beam reconstruction operators and end-to-end learning.