April 2010
Volume 51, Issue 13
Free
ARVO Annual Meeting Abstract  |   April 2010
Wavelet Denoising of Multiple-Frame OCT Data Enhanced by a Correlation Analysis
Author Affiliations & Notes
  • M. A. Mayer
    Pattern Recognition Lab, University of Erlangen-Nuremberg, Erlangen, Germany
    School of Advanced Optical Technologies (SAOT), University of Erlangen-Nuremberg, Germany
  • M. Wagner
    Pattern Recognition Lab, University of Erlangen-Nuremberg, Erlangen, Germany
  • J. Hornegger
    Pattern Recognition Lab, University of Erlangen-Nuremberg, Erlangen, Germany
    School of Advanced Optical Technologies (SAOT), University of Erlangen-Nuremberg, Germany
  • R. P. Tornow
    Department of Ophthalmology, University of Erlangen-Nuremberg, Germany
  • Footnotes
    Commercial Relationships  M.A. Mayer, None; M. Wagner, None; J. Hornegger, None; R.P. Tornow, None.
  • Footnotes
    Support  Erlangen Graduate School in Advanced Optical Technologies (SAOT)
Investigative Ophthalmology & Visual Science April 2010, Vol.51, 1777. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      M. A. Mayer, M. Wagner, J. Hornegger, R. P. Tornow; Wavelet Denoising of Multiple-Frame OCT Data Enhanced by a Correlation Analysis. Invest. Ophthalmol. Vis. Sci. 2010;51(13):1777.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract
 
Purpose:
 

Speckle noise suppression on OCT images is currently performed by averaging multiple frames. In contrast to this common approach we propose a novel wavelet based method that uses the structural properties of the actual image content to better differentiate between speckle and relevant tissue information.

 
Methods:
 

Each of the recorded single B-Scans is decomposed by a wavelet analysis. The wavelet coefficients representing a local frequency analysis of the images are then adapted by weights computed out of a statistical analysis. For each coefficient on every frame individual weights are calculated. They are constructed out of the inter frame intensity differences and correlations in small neighborhoods. Finally the modified single frames are wavelet reconstructed and averaged.To test the algorithm 455 linear B-scans were aquired from a pig's eye ex vivo with a Spectralis HRA+OCT, Heidelberg Engineering. Correlated noise was avoided by slightly moving the eye every 13 frames. All images are rigidly registered and averaged to form a noise suppressed gold standard. The signal-to-noise ratio (SNR) as well as sharpness reduction at selected borders using Full-Width-Half-Maximum is measured.

 
Results:
 

Using the proposed method with only 8 frames we achieve a similar result compared to common averaging of 35 frames. This improvement is verified by an evaluation on multiple randomly selected sets. The SNR is increased by 104.0% while sharpness is only slightly reduced by 9.7%. Figure 1 shows the mean of 8 frames (left) compared to the output of our algorithm (right).

 
Conclusions:
 

The proposed method makes use of the spatially varrying noise statistics of multiple frame OCT-data of the same region. The evaluation showed that the SNR of OCT scans can be significantly improved while edge sharpness remains. The algorithm is applicable in 3D and can therefore be used to acquire high-quality OCT volume data in a shorter amount of time.  

 
Keywords: image processing • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) 
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×