April 2014
Volume 55, Issue 13
Free
ARVO Annual Meeting Abstract  |   April 2014
Retinal tissue discrimination in an animal model using hyperspectral computed tomographic imaging spectroscopy (HCTIS)
Author Affiliations & Notes
  • Amir H Kashani
    Ophthalmology, University of Southern California, Los Angeles, CA
  • Mark Wong
    Goddard Space Flight Center, NASA, Greenbelt, MD
  • Mark S Humayun
    Ophthalmology, University of Southern California, Los Angeles, CA
  • Footnotes
    Commercial Relationships Amir Kashani, Reichert Technologies (F); Mark Wong, None; Mark Humayun, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science April 2014, Vol.55, 2080. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Amir H Kashani, Mark Wong, Mark S Humayun; Retinal tissue discrimination in an animal model using hyperspectral computed tomographic imaging spectroscopy (HCTIS). Invest. Ophthalmol. Vis. Sci. 2014;55(13):2080.

      Download citation file:


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

      ×
  • Supplements
Abstract
 
Purpose
 

To show that HCTIS can provide highly coregistered temporal, spatial and spectral data with sufficient resolution to allow reconstruction and spectral differentiation of the vascular tree.

 
Methods
 

Two rabbits (3-4kg) were anesthetized and dilated using 2.5% phenylephrine and 0.5% tropicamide ophthalmic solutions. Duplicate color images of the fundus were acquired with a custom made HCTIS device coupled to a standard, commercially available fundus camera as described before (Kashani AH et al., PLoS One 2011). Each image contains up to 76 spectral bands and was obtained using standard fundus photography. A previously described unsupervised method was used to reconstruct hyperspectral images using an iterative algorithm (Kashani AH et al., PLoS One 2011). Spectral similarity was assessed using custom made non-supervised algorithms (based on vector analysis using spectral similarity mapping algorithms) to compare the spectral similarity of tissues within each voxel of the hyperspectral dataset and create a pseudocolored similarity map of the image. Similarity maps of the tissue were made using reference voxels manually selected from the retinal vasculature and the neural tissue as representative tissue. Analysis of two representative images from each rabbit was performed.

 
Results
 

Figure 1 illustrates representative pseudocolored images displaying the relative similarity of various voxels from the rabbit retina (red = 100% similarity and blue = 0% similarity). When reference voxels were from non-vascular tissue (Figure 1A) similarity maps successfully identified non-vascular tissue components throughout the target image with high accuracy. When reference voxels were from the vascular tree (Figure 1B) the similarity mapping algorithm successfully identified a large portion of the vascular tree without any supervision. Similar results were obtained for all images analysed.

 
Conclusions
 

HCTIS images provide highly coregistered temporal, spatial and spectral data that can be used with unsupervised spectral similarity algorithms to identify relevant target tissues. These mapping methods may be useful in identifying various kinds of tissues in health and disease.

     
Keywords: 552 imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • 551 imaging/image analysis: non-clinical  
×
×

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.

×