Peter Carr

Tamir Yedidya


Ph.D. Candidate
Computer Vision and Robotics Group
Research School of Information Sciences and Engineering
Australian National University (ANU)

NICTA (National ICT Australia)
7 London Circuit
Canberra, ACT 2601
Australia
Email:tamir.yedidya at anu.edu.au

Winning photos at the ANU Photography Calendar competition:
2007 cheering
2008 summer hail storm (a personal favorite)
2008 Winners welcome

Research Interests
I am currently working on the automatic detection of signs related to dry eyes, under the supervision of Prof. Richard Hartley. My research involves applying computer vision and image processing methods to eye images.
We invented a number of new algorithms to automatically detect various dryness signs:
  • We detect dry areas in fluorescein videos of the anterior of the eye, which are captured using a portable camera.
  • We produce a novel segmentation result called dryness image (see example below), which depicts the various degrees of tear film thinning.
  • We define a new clinical value called Digital Electronic Break Up Time (DEBUT), which is robust and invariant to eye movements, illumination changes and operator.
  • We explain how monotonic constraints can be applied to dry eye detection and other imaging modalities.
  • We evaluate the tear meniscus height and shape using graph-cuts.

Dry Eye Syndrome

 There have been very little computer-vision research related to corneal images (images of the front of the eye) and the associated diseases, such as the Dry Eye Syndrome. Yet, there are almost 5 million Americans over 50 years old that suffer from dry eye. It is possible that even you (yes you..) suffer from dry eye. Have you ever had an uneasy feeling in your eyes after using the computer for a couple of hours? Maybe, you had to take a break or it triggered a headache? Watching the computer, wearing contact lens and other daily actions can all cause dry eye. As dry eye affects us in our everyday activities, it can reduces drastically the quality of life. 		An example for fluorescein video
Dryness Image
Tear film break and DEBUT


An example for a recorded video. Watch how dryness regions form simultaneously at various areas
time t = 0 seconds time t = 4 seconds time t = 8 seconds time t = 12 seconds
Still images taken from a video, demonstrating the evolution of dry areas over time.

Recently (2007) a special dry eye committee (DEWS) produced a report summarizing what is currently known regarding dry eye. The committe mentions that no single diagnostic test can be performed to reliably distinguish individuals with and without dry eye. The committee concludes that no "gold standard" exists for the diagnosis of dry eye and new diagnostic tests should be developed. Here is where we come for aid..

Our Dry Eye Analysis Tool

 Our automatic system is based on the Fluorescein Break Up Time (FBUT) test (see Figure above), which is one of the most commonly used tests by the clinicians. However, this test is highly subjective ( see an example ) and has a number of additional limitations. We confront these challenging tasks and define a robust measurement to estimate dryness which is repeatable, not subjective and not operator-dependent. In addition, we build a system which is capable of simultaneously producing results of a few diagnostic criteria related to dry eye.

System Benefits

Algorithmic Difficutlies



Portable camera using tear film module The dryness image produced for the video images above. The colored areas are related to regions of break. The other areas are regions of tear film thinning - the brighter the dryer it became


Research Activities

 Resume