July 2019
Volume 60, Issue 9
Free
ARVO Annual Meeting Abstract  |   July 2019
Robotically Aligned OCT Scanner for Automated Patient Tracking Retinal Imaging
Author Affiliations & Notes
  • Mark Draelos
    Biomedical Engineering, Duke University, Durham, North Carolina, United States
  • Pablo Ortiz
    Biomedical Engineering, Duke University, Durham, North Carolina, United States
  • Ruobing Qian
    Biomedical Engineering, Duke University, Durham, North Carolina, United States
  • Christian Viehland
    Biomedical Engineering, Duke University, Durham, North Carolina, United States
  • Kris Hauser
    Eletrical and Computer Engineering, Duke University, North Carolina, United States
  • Anthony N Kuo
    Duke University Medical Center, Ophthalmology, North Carolina, United States
  • Joseph A Izatt
    Biomedical Engineering, Duke University, Durham, North Carolina, United States
  • Footnotes
    Commercial Relationships   Mark Draelos, None; Pablo Ortiz, None; Ruobing Qian, None; Christian Viehland, None; Kris Hauser, None; Anthony Kuo, Leica Microsystems (P); Joseph Izatt, Carl Zeiss Meditec (P), Carl Zeiss Meditec (R), Leica Microsystems (P), Leica Microsystems (R)
  • Footnotes
    Support  NIH F30-EY027280, NIH R01- EY029302, Coulter Translational Partnership
Investigative Ophthalmology & Visual Science July 2019, Vol.60, 1268. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Mark Draelos, Pablo Ortiz, Ruobing Qian, Christian Viehland, Kris Hauser, Anthony N Kuo, Joseph A Izatt; Robotically Aligned OCT Scanner for Automated Patient Tracking Retinal Imaging. Invest. Ophthalmol. Vis. Sci. 2019;60(9):1268.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : Current clinical optical coherence tomography (OCT) systems require stabilization of patient heads, usually with a chin and head rest, to reduce motion artifacts in the images. This limits the use of these OCT systems for those who cannot be stabilized in this way, such as unconscious, bedbound, posturally-limited, and very young patients. Using a robot-mounted retinal OCT scanner with face and eye tracking capabilities, we demonstrate high quality OCT imaging of freestanding subjects’ eyes.

Methods : Our automatic OCT system consisted of a custom retinal scanner (Fig. 1) mounted on a 6 degree-of-freedom robot arm (UR3, Universal Robotics), a custom 100 kHz swept-source OCT engine with a motorized reference arm, and two fixed face tracking RGB-D cameras (RealSense D415, Intel). The retinal scanner had a 100 mm working distance for subject comfort, a 16-degree retinal field of view, and a fast steering mirror (FSM) (OIM202.3, Optics In Motion) at the retinal conjugate plane for pupil pivot aiming. An inline pupil camera shared the OCT objective to provide lateral pupil tracking via the FSM whereas an offset pupil camera formed a stereo pair for axial pupil tracking via the motorized reference arm. The face tracking cameras detected facial landmarks using OpenFace 2.0 to guide alignment until the pupil entered the scanner’s tracking field of view. Pupil tracking, face tracking, and robot control operated at 250 Hz, 30 Hz, and 125 Hz, respectively. We evaluated the system by imaging a model eye (Rowe Technical Design) embedded in a face phantom as well as a human subject consented under an IRB-approved protocol. We corrected residual axial motion by registration to a single orthogonal B-scan through the fovea.

Results : Model eye imaging showed tracking with 8.8 Hz bandwidth, 15 µm and 25 µm accuracy and precision laterally, and 110 µm and 480 µm accuracy and precision axially. Human subject imaging yielded OCT volumes suitable for retinal evaluation (Fig. 2) with 5 seconds of alignment time and 2.5 seconds of imaging time at 200x800x1376 resolution for a single eye. These OCT volumes demonstrated suppression of lateral motion.

Conclusions : Active robot tracking and scan aiming enables OCT imaging in freestanding subjects.

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

 

Robot-mounted OCT scanner aligned to a subject’s left eye.

Robot-mounted OCT scanner aligned to a subject’s left eye.

 

Example tracked OCT en face projection (top), B-scan (middle), and registered volumetric render (bottom) showing a human subject’s macula.

Example tracked OCT en face projection (top), B-scan (middle), and registered volumetric render (bottom) showing a human subject’s macula.

×
×

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.

×