June 2013
Volume 54, Issue 15
Free
ARVO Annual Meeting Abstract  |   June 2013
Evaluation of a New Registration Algorithm Using iVue SD-OCT
Author Affiliations & Notes
  • Jing Cui
    Optovue, Inc, Fremont, CA
  • Ben Jang
    Optovue, Inc, Fremont, CA
Investigative Ophthalmology & Visual Science June 2013, Vol.54, 5515. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Jing Cui, Ben Jang; Evaluation of a New Registration Algorithm Using iVue SD-OCT. Invest. Ophthalmol. Vis. Sci. 2013;54(15):5515.

      Download citation file:


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

      ×
  • Supplements
Abstract
 
Purpose
 

Registration is essential to align images generated at different visits. Blood vessel patterns are often extracted for this purpose. However, blood vessel extraction is challenging due to scan quality or pathologies. This study is to evaluate the performance of a new registration algorithm using iVue spectral domain (SD) optical coherence tomography (OCT) system (Optovue, Fremont, CA).

 
Methods
 

The algorithm first extracts an edge map based on image gradient magnitude and orientation. A new adaptive thresholding technique is applied to the edge map to obtain a blood vessel map. Affine transformation is determined based on the blood vessel maps of the two images to be registered. The algorithm is capable of detecting fine blood vessels but insensitive to scatter noise, non-uniform background, or artifacts in images. To evaluate the new registration algorithm, iVue SD-OCT “Retina Map” scan pattern was used to acquire data in a 6mmx6mm square region centered at fovea. It generates 20 B-scan and one en-face OCT images. Retina thickness map is computed based on the automated segmentation of retinal layers on B-scan images. Retina thickness maps generated at different visits of the same eye can be aligned by registering associated en-face images. The data was collected from 30 subjects with informed consent. Each patient was scanned on one or both eyes. A total of 46 eyes were scanned, including 33 normal eyes and 13 eyes with various retinal pathologies. Each tested eye has two or more “Retina Map” scans, in which the en-face image of the first scan was used as reference and registered with the en-face images of the other scans of the same eye. Totally, 58 pairs of registration were performed. Each pair of registration was evaluated by calculating the ratio of the number of overlapped pixels to the total number of pixels in the blood vessel maps. If the ratio was greater than 50%, the registration was graded as “successful”, otherwise it was graded as “fail”.

 
Results
 

For the total 58 registration pairs, 49 were graded as “successful” (as shown in Fig. 1) and 9 were graded as “fail”. The 9 failure cases were all due to the very low scan quality in which the blood vessels are barely identifiable.

 
Conclusions
 

The new registration algorithm showed improved performance of registering data in iVue SD-OCT system.

  
Keywords: 688 retina • 549 image processing  
×
×

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.

×