Progressive self-supervised blind-spot denoising method for LDCT denoising
Yichao Liu, Yueyang Teng, Junwen Guo
TL;DR
The paper tackles LDCT denoising without requiring paired NDCT data by introducing a progressive self-supervised blind-spot denoising framework. It combines step-wise masking with a denoising horizon of $k$ steps, a $J$-invariant loss, and noise-based regularization via $n_1$ and $n_2$ added to inputs and targets. On the AAPM LDCT Grand Challenge dataset, the method consistently outperforms existing self-supervised approaches and is competitive with supervised denoisers such as RED-CNN and Cycle-GAN, with performance robust across patients. The approach is architecture-agnostic, enabling integration with different backbones and offering a practical, data-efficient solution for LDCT denoising in clinical workflows.
Abstract
Self-supervised learning is increasingly investigated for low-dose computed tomography (LDCT) image denoising, as it alleviates the dependence on paired normal-dose CT (NDCT) data, which are often difficult to acquire in clinical practice. In this paper, we propose a novel self-supervised training strategy that relies exclusively on LDCT images. We introduce a step-wise blind-spot denoising mechanism that enforces conditional independence in a progressive manner, enabling more fine-grained denoising learning. In addition, we add Gaussian noise to LDCT images, which acts as a regularization and mitigates overfitting. Extensive experiments on the Mayo LDCT dataset demonstrate that the proposed method consistently outperforms existing self-supervised approaches and achieves performance comparable to, or better than, several representative supervised denoising methods.