September 2019
Volume 19, Issue 10
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2019
Perceptual Learning of Optical Coherence Tomography Image Classification
Author Affiliations & Notes
  • Evan M Palmer
    Department of Psychology, San Jose State University
  • Elnaz Amiri
    Industrial Systems Engineering, San Jose State University
    Carl Zeiss Meditec, Inc., Dublin, CA, United States
  • Patty Sha
    Carl Zeiss Meditec, Inc., Dublin, CA, United States
  • Sophia Yu
    Carl Zeiss Meditec, Inc., Dublin, CA, United States
  • Gregory Anderson
    Carl Zeiss Meditec, Inc., Dublin, CA, United States
  • Gary C Lee
    Carl Zeiss Meditec, Inc., Dublin, CA, United States
Journal of Vision September 2019, Vol.19, 28. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Evan M Palmer, Elnaz Amiri, Patty Sha, Sophia Yu, Gregory Anderson, Gary C Lee; Perceptual Learning of Optical Coherence Tomography Image Classification. Journal of Vision 2019;19(10):28.

      Download citation file:

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

  • Supplements

Purpose: This study explored how perceptual learning methods may be applied to improve diagnosis of age-related macular degeneration (AMD) in optical coherence tomography (OCT) retinal scan images. Perceptual learning can occur when observers perform rapid image classification tasks with immediate feedback. This study is a first step towards developing software for training optometry students and technicians. Methods: The 189 images used for this study were taken on CIRRUS 5000 HD-OCT (ZEISS, Dublin, CA). Each image was a B-scan selected from 6mm × 6mm cubes of the macula. Twenty participants with no prior training in optometry or medical image diagnosis completed the study. Images were presented on 23” Dell P2317H monitors driven by 1.4 GHz Mac Mini computers running MATLAB (MathWorks, Natick, MA) software with the Psychophysics toolbox (Brainard, 1997; Pelli, 1997). Participants classified each image as either wet or dry AMD via keypress. Classification accuracy and response times were collected pre and post a 15 minute perceptual training session. Accuracy feedback was provided only during the training session. Response times and accuracy of classifications during the pre- and post-test periods were assessed via analyses of variance. Results: Analyses detected a main effect of response time, p = .002, such that correct classifications were faster in the post-test than the pre-test. A main effect of image type indicated that observers were more accurate at classifying wet than dry images, p < .001. Finally, there was an interaction in the accuracy data, such that accuracy improved for wet AMD images but not for dry AMD images, p = .034. Conclusions: These preliminary results suggest that it is possible to improve the speed and accuracy of wet vs. dry AMD classification by novices with a 15 minute training session. We are continuing to collect data to improve our AMD classification training methods.


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.