Clustering-Oriented Multiple Convolutional Neural Networks for Optical Coherence Tomography Image Denoising

  • Xiangkai Wei
  • , Xiaoming Liu*
  • , Aihui Yu
  • , Tianyu Fu
  • , Dong Liu
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The speckle noise is an inherent coproduct of OCT imaging that is a significant direct influence factor of image quality, thus OCT image denoising is needed. Most existing OCT image denoising methods usually use only part of the priori information of the OCT image, but neglect the change of the texture, structure and other features of the OCT image. To address this, we introduce a framework for OCT image denoising by multiple CNNs based on clustering and residual learning. Our proposed method not only utilizes the automatic feature learning ability of CNNs but also adapts them to depict diversity of noise characteristics in different areas of noisy input. Our framework achieves great visual and quantitative performance.

Original languageEnglish
Title of host publicationProceedings - 2018 11th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics, CISP-BMEI 2018
EditorsWei Li, Qingli Li, Lipo Wang
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781538676042
DOIs
Publication statusPublished - 2 Jul 2018
Externally publishedYes
Event11th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics, CISP-BMEI 2018 - Beijing, China
Duration: 13 Oct 201815 Oct 2018

Publication series

NameProceedings - 2018 11th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics, CISP-BMEI 2018

Conference

Conference11th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics, CISP-BMEI 2018
Country/TerritoryChina
CityBeijing
Period13/10/1815/10/18

Keywords

  • CNNs
  • OCT
  • clustering
  • denoising
  • residual learning

Fingerprint

Dive into the research topics of 'Clustering-Oriented Multiple Convolutional Neural Networks for Optical Coherence Tomography Image Denoising'. Together they form a unique fingerprint.

Cite this