Table 1 Overview of the literature on OCT image denoising.
From: Self-supervised model-informed deep learning for low-SNR SS-OCT domain transformation
Data-driven methods | Paired data | DesPecNet 8 | Network-based denoiser with knowing noise level |
Gour et al. 9 | Removing speckle noise using supervised trained residual network. | ||
Devella et al. 10 | Train a denoiser network on averaged frames. | ||
Halupka et al. 11 | Train a denoiser network on averaged frames. | ||
Unpaired data | Wang et al. 14 | Generative-based image-to-image translation. | |
Guo et al. 15 | Generative network for image-to-image translation. | ||
Ma et al. 16 | Generative network for image-to-image translation. | ||
Yu et al. 17 | Generative network of image-to-image translation | ||
Model-driven methods | Filtering | BM3D 18 | Non-local patch-wise image modeling. |
Wavelet-based dictionary learning 20 | Wavelet based dictionary learning for image representation. | ||
Statistical | SDE decomposition 21 | Stochastic differential equation-based model to whitening OCT images. | |
α-stable model | Symmetric α-stable distribution modeling on whitening transformation. | ||
BKF model | Modeling a patch-based Gaussian transform based. | ||
Hybrid methods | Iterative models | Diffusion probabilistic model 26 | Diffusion probabilistic model for image generator as prior inside an iterative optimization. |
Score based model 27 | Score-based model as denoiser model. | ||
Single iteration model | B2U 29 | Noise2Void model for OCT denoising. | |
N2N 25 | Noise2Noise model for OCT denoising. |
