July 2020
Volume 61, Issue 9
Free
ARVO Imaging in the Eye Conference Abstract  |   July 2020
Robust detection of mirror image artifacts using full-range reconstruction of OCT
Author Affiliations & Notes
  • Yuanzhi Liu
    Carl Zeiss Meditec, Inc., Dublin, California, United States
  • Homayoun Bagherinia
    Carl Zeiss Meditec, Inc., Dublin, California, United States
  • Ali Fard
    Carl Zeiss Meditec, Inc., Dublin, California, United States
  • Hugang Ren
    Carl Zeiss Meditec, Inc., Dublin, California, United States
  • Footnotes
    Commercial Relationships   Yuanzhi Liu, Carl Zeiss Meditec, Inc. (E); Homayoun Bagherinia, Carl Zeiss Meditec, Inc. (E); Ali Fard, Carl Zeiss Meditec, Inc. (E); Hugang Ren, Carl Zeiss Meditec, Inc. (E)
  • Footnotes
    Support  None
Investigative Ophthalmology & Visual Science July 2020, Vol.61, PB0088. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Yuanzhi Liu, Homayoun Bagherinia, Ali Fard, Hugang Ren; Robust detection of mirror image artifacts using full-range reconstruction of OCT. Invest. Ophthalmol. Vis. Sci. 2020;61(9):PB0088.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : In Fourier domain optical coherence tomography (OCT), the reconstruction results in two symmetric images of the sample structure around the 0-delay time and only the half-range is usually cropped for analysis and display. The mirror-image (MI) introduces artifacts to the real-image (RI) if they are not well separated. Hence a local MI artifact detection is essential for robust scan quality assessment, z-tracking, and en face analysis.

Methods : An algorithm utilizing all the information of the full-range OCT reconstruction is developed to detect the MI artifact. Basically, RI becomes sharper than MI after the numerical dispersion compensation when there is unbalanced dispersion between the sample and reference arm. The 1st lag of the autocorrelation function (ACF) of an A-scan in RI is therefore smaller. This property is utilized as the indicator for automatically identifying RI and MI.

Results : A typical full-range OCT reconstruction acquired from CIRRUSTM 6000 (ZEISS, Dublin, CA) is shown in Fig. 1(a), where RI appears in the bottom half region (common display and processing window) with a green box and MI is in the top half region (red box). Fig. 1(b) shows the separated ACFs for these two regions along the dash line in Fig.1 (a) where RI has a lower value of the 1st lag of the ACF. Figs. 2(a)-(c) and (d)-(f) show the capability of this algorithm robustly detecting MI in varied cases: RI (in the green window) and MI are separated; MI partially appears in the green window; and MI fully enters the green window while RI is in the red window. Figs. 2(g)-(i) and (h)-(j) indicate even the RI and MI has partial overlap (ONH region) they can still be detected. Fig. 2(k) shows an application of using this method to automatically discard the contaminated images before implementing further analysis. An en face binary map is generated by comparing the ACFs along each A-line between the red and green windows. If the ACF is higher in the green window than the red one, the pixel in the binary map is set to 1 (contaminated), otherwise is 0.

Conclusions : We proposed a new robust technique for fast detection of RI and MI in OCT by comparing their 1st lags of ACF. This method can be easily implemented in the existing processing pipeline of various Fourier domain (spectral domain and swept-source) OCT systems.

This is a 2020 Imaging in the Eye Conference abstract.

 

Figure 1. Full-range OCT reconstruction and the ACF of one A-scan.

Figure 1. Full-range OCT reconstruction and the ACF of one A-scan.

 

Figure 2. Robust detection of mirror image artifacts.

Figure 2. Robust detection of mirror image artifacts.

×
×

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.

×