Autoencoder-based dense denoiser and block-based wiener filter for noise reduction of optical coherence tomography

Optical Coherence Tomography (OCT) is an advanced imaging modality used for diagnosis of retinal abnormalities. OCT is acquired using low coherence light waves, typically infra-red waves having resolution in micrometres so as to capture the retinal layers present in the eye. Analysing variation in thickness of different retinal layers using OCT can be used for diagnosis. However, these layers are not clearly visible due to the presence of varying amounts of speckle noise, due to which the efficacy of further diagnosis gets compromised. Despite multiple approaches being available for denoising of OCT images, an undesirable over smoothening of images, leads to loss of structural edge details, thereby leading to inaccurate diagnosis. Thus, an efficient approach that removes speckle noise, without compromising on the significant image details, is preferred. This paper presents an approach to eliminate the speckle noise from OCT images using an Autoencoder-based Dense Denoiser (ADD) neural network and Block-based Wiener Filter (BBWF).

Automated summary

Optical Coherence Tomography (OCT) is an advanced imaging technique used for diagnosing retinal abnormalities by capturing the retinal layers using low coherence light waves. However, the presence of speckle noise hinders accurate diagnosis. This paper proposes an approach using an Autoencoder-based Dense Denoiser (ADD) neural network and Block-based Wiener Filter (BBWF) to eliminate speckle noise without compromising important image details.