April 2011
Volume 52, Issue 14
Free
ARVO Annual Meeting Abstract  |   April 2011
A New Technique to Discriminate Glaucomatous Eyes Based on Super Pixel Machine Classifier Analysis of Three-Dimensional (3D) Spectral Domain Optical Coherence Tomography (SD-OCT) Images
Author Affiliations & Notes
  • Juan Xu
    UPMC Eye Center, Eye & Ear Institute, Ophthalmology & Visual Science Research Center, Dept. of Ophthalmology, Univ. of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
  • Hiroshi Ishikawa
    UPMC Eye Center, Eye & Ear Institute, Ophthalmology & Visual Science Research Center, Dept. of Ophthalmology, Univ. of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    Dept. of Bioengineering, Swanson School of Engineering, Univ. of Pittsburgh, Pittsburgh, Pennsylvania
  • Gadi Wollstein
    UPMC Eye Center, Eye & Ear Institute, Ophthalmology & Visual Science Research Center, Dept. of Ophthalmology, Univ. of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
  • Richard A. Bilonick
    UPMC Eye Center, Eye & Ear Institute, Ophthalmology & Visual Science Research Center, Dept. of Ophthalmology, Univ. of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    Dept. of Biostatistics, Graduate School of Public Health, Univ. of Pittsburgh, Pittsburgh, Pennsylvania
  • Lindsey S. Folio
    UPMC Eye Center, Eye & Ear Institute, Ophthalmology & Visual Science Research Center, Dept. of Ophthalmology, Univ. of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    Dept. of Bioengineering, Swanson School of Engineering, Univ. of Pittsburgh, Pittsburgh, Pennsylvania
  • Michelle G. Sandrian
    UPMC Eye Center, Eye & Ear Institute, Ophthalmology & Visual Science Research Center, Dept. of Ophthalmology, Univ. of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    Dept. of Bioengineering, Swanson School of Engineering, Univ. of Pittsburgh, Pittsburgh, Pennsylvania
  • Larry Kagemann
    UPMC Eye Center, Eye & Ear Institute, Ophthalmology & Visual Science Research Center, Dept. of Ophthalmology, Univ. of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    Dept. of Bioengineering, Swanson School of Engineering, Univ. of Pittsburgh, Pittsburgh, Pennsylvania
  • Joel S. Schuman
    UPMC Eye Center, Eye & Ear Institute, Ophthalmology & Visual Science Research Center, Dept. of Ophthalmology, Univ. of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    Dept. of Bioengineering, Swanson School of Engineering, Univ. of Pittsburgh, Pittsburgh, Pennsylvania
  • Footnotes
    Commercial Relationships  Juan Xu, Bioptigen, Inc. (P); Hiroshi Ishikawa, Bioptigen, Inc. (P); Gadi Wollstein, Bioptigen, Inc. (P), Optovue (F); Richard A. Bilonick, None; Lindsey S. Folio, None; Michelle G. Sandrian, None; Larry Kagemann, None; Joel S. Schuman, Bioptigen, Inc. (P), Carl Zeiss Meditec, Inc. (P)
  • Footnotes
    Support  NIH R01-EY013178; P30-EY008098; Eye and Ear Foundation (Pittsburgh, PA); Research to Prevent Blindness.
Investigative Ophthalmology & Visual Science April 2011, Vol.52, 3960. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Juan Xu, Hiroshi Ishikawa, Gadi Wollstein, Richard A. Bilonick, Lindsey S. Folio, Michelle G. Sandrian, Larry Kagemann, Joel S. Schuman; A New Technique to Discriminate Glaucomatous Eyes Based on Super Pixel Machine Classifier Analysis of Three-Dimensional (3D) Spectral Domain Optical Coherence Tomography (SD-OCT) Images. Invest. Ophthalmol. Vis. Sci. 2011;52(14):3960.

      Download citation file:


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

      ×
  • Supplements
Abstract
 
Purpose:
 

To develop a 3D SD-OCT data analysis technique using super pixel machine classifier to improve glaucoma discrimination.

 
Methods:
 

192 eyes of 96 subjects (44 normal, 59 glaucoma suspect and 89 glaucomatous eyes) were scanned with 3D SD-OCT (Cirrus HD-OCT; Carl Zeiss Meditec, Inc., Dublin, CA; ONH Cube 200x200 scan protocol). 400 super pixels were automatically segmented on 2D retinal nerve fiber layer (RNFL) thickness maps by grouping homogeneous neighbor pixels using our custom software (Figure). The RNFL thickness maps were adjusted by multiplying RNFL relative reflectivity. Super pixel features, i.e., average intensity value and size were used as inputs to 3 machine classifiers (boosting learning, support vector machine, and k-nearest neighbor) to automatically identify abnormal eyes. 10 fold cross validation area under the receiver operating characteristics (AUC) of the machine classifier outputs, for discriminating between normal and glaucomatous eyes, were compared with the circumpapillary RNFL thickness generated by Cirrus HD-OCT software.

 
Results:
 

Boosting learning method obtained the best AUCs amongst the machine classifiers, where the AUC of normal vs glaucoma suspect eyes was statistically significantly improved from 0.677 (the circumpapillary RNFL thickness performance) to 0.829 (p=0.018, DeLong test; Table). The AUCs did not show significant differences for all other comparisons.

 
Conclusions:
 

A novel 3D OCT analysis technique was better at discriminating between normal and glaucoma suspect eyes than the traditional circumpapillary RNFL analysis, and performed similarly for normal vs glaucomatous eyes. This new method has the potential to improve early detection of glaucomatous damage.  

 
Keywords: image processing • imaging/image analysis: clinical • 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.

×