July 2019
Volume 60, Issue 9
Open Access
ARVO Annual Meeting Abstract  |   July 2019
Corneal Confocal Image Fusion
Author Affiliations & Notes
  • Simone Pajaro
    NIDEK Technologies Srl, Albignasego, PD, Italy
  • Giulia Menin
    Department of Information Engineering, University of Padova, Padova, PD, Italy
  • Michele Pascolini
    NIDEK Technologies Srl, Albignasego, PD, Italy
  • Footnotes
    Commercial Relationships   Simone Pajaro, NIDEK Technologies Srl (E); Giulia Menin, None; Michele Pascolini, NIDEK Technologies Srl (E)
  • Footnotes
    Support  None
Investigative Ophthalmology & Visual Science July 2019, Vol.60, 165. doi:https://doi.org/
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Simone Pajaro, Giulia Menin, Michele Pascolini; Corneal Confocal Image Fusion. Invest. Ophthalmol. Vis. Sci. 2019;60(9):165. doi: https://doi.org/.

      Download citation file:

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

  • Supplements

Purpose : Image fusion is becoming every day more popular as a technique for augmenting the information carried by a single representation using the one collected from multiple sources. The purpose of this work is to study a possible approach for composing a sequence of corneal images acquired with a new noncontact confocal microscope developed at NIDEK Technologies Srl.

Methods : A prototype of a confocal microscope based on a point-scanning laser working at 640nm coupled with an avalanche photodiode (APD) is used to collect slices of the cornea. Collected gray-scale images have a size of 1024x1024 pixels and 8-bit depth. The optical system makes use of a 20x lens and permits to observe a corneal area of 400x400 μm2. This area is comparable with the one of other commercial products as the NIDEK Confoscan 4 or the Heidelberg HRT 3 Rostock Cornea module. For each sequence of collected images from the same eye, a stack of images is formed respectively for the stroma, epithelium, and endothelium.
Each stack of images consists of about 5-6 images and is processed as follows. An initial pre-processing reduces the noise of the acquired data. Images are then registered in order to compensate for motion occurred during the acquisition. The resulting image is successively filtered with a moving-average spatial filter. The stack of registered images is then processed with a majority filter based on pixel intensities in order to produce a depth map containing for each pixel the slice having the best focus for that pixel. Finally, a composite image is created from the stack by taking pixels from the best-focus images as indicated by the depth map.

Results : The algorithm was tested on one stack of 5 images of the epithelium, one stack of 6 images for the stroma, and one stack of 6 images of the endothelium of enucleated pig eyes. Composite images of these stacks showed that the processing algorithm works as expected, by enhancing the overall focus and anatomical details. Example of an image stack and the corresponding reconstruction is given in figures 1 and 2.

Conclusions : The proposed approach to the problem of image fusion permits a new way of extending the depth of focus of corneal confocal images by applying image registration and a majority filter. Results are encouraging and open the way for a refinement of the proposed ideas.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.


Figure 1: Stack of images of endothelium from an enucleated pig eye

Figure 1: Stack of images of endothelium from an enucleated pig eye


Figure 2: (a) Fusion map and (b) resulting “fused” image

Figure 2: (a) Fusion map and (b) resulting “fused” image


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.