Semantic denoising autoencoders for retinal optical coherence tomography

Max-Heinrich Laves; Sontje Ihler; Lüder Alexander Kahrs; Tobias Ortmaier

doi:10.1117/12.2526936

19 July 2019 Semantic denoising autoencoders for retinal optical coherence tomography

Max-Heinrich Laves, Sontje Ihler, Lüder Alexander Kahrs, Tobias Ortmaier

Proceedings Volume 11078, Optical Coherence Imaging Techniques and Imaging in Scattering Media III; 1107818 (2019) https://doi.org/10.1117/12.2526936
Event: European Conferences on Biomedical Optics, 2019, Munich, Germany

Abstract

Noise in speckle-prone optical coherence tomography tends to obfuscate important details necessary for medical diagnosis. In this paper, a denoising approach that preserves disease characteristics on retinal optical coherence tomography images in ophthalmology is presented. We propose semantic denoising autoencoders, which combine a convolutional denoising autoencoder with a priorly trained ResNet image classifier as regularizer during training. This promotes the perceptibility of delicate details in the denoised images that are important for diagnosis and filters out only informationless background noise. With our approach, higher peak signal-to-noise ratios with PSNR = 31.0 dB and higher classification performance of F₁ = 0.92 can be achieved for denoised images compared to state-of-the-art denoising. It is shown that semantically regularized autoencoders are capable of denoising retinal OCT images without blurring details of diseases.

Citation Download Citation

Max-Heinrich Laves, Sontje Ihler, Lüder Alexander Kahrs, and Tobias Ortmaier "Semantic denoising autoencoders for retinal optical coherence tomography", Proc. SPIE 11078, Optical Coherence Imaging Techniques and Imaging in Scattering Media III, 1107818 (19 July 2019); https://doi.org/10.1117/12.2526936

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available