Author Topic: Composite Eye - defining pupil  (Read 6064 times)

dmw69

  • Newbie
  • *
  • Posts: 25
    • View Profile
Composite Eye - defining pupil
« on: May 20, 2013, 07:21:20 pm »
I have started using the composite eye function. I use this to get greater areas of coverage by getting the patient to fixate up, down, L, R; and also for greater accuracy by combining four maps in primary gaze.

In the instructions (p51) it says to ensure the pupil was correctly detected in each exam, before creating the composite eye.

Sometimes the detected pupil is only correct for say half the circumference, and completely wrong in the other half. Since re-defining the pupil margin on 4 or 5 exams for each eye can be time consuming, is it necessary to define the pupil on *all* exams that are to be combined, or only on one (which will end up becoming the "most central exam")?

Also, how closely does the redefined pupil have to align with the actual pupil (is it OK as long as it's concentric?); and does the software ignore obviously non-circular (mis-detected) pupils?

dmw69

  • Newbie
  • *
  • Posts: 25
    • View Profile
Re: Composite Eye - defining pupil
« Reply #1 on: September 18, 2013, 04:35:53 pm »
bump

Michael

  • Senior Software Engineer
  • Administrator
  • Hero Member
  • *****
  • Posts: 10149
    • View Profile
Re: Composite Eye - defining pupil
« Reply #2 on: September 23, 2013, 04:46:05 pm »
Hi,

Quote
Sometimes the detected pupil is only correct for say half the circumference, and completely wrong in the other half. Since re-defining the pupil margin on 4 or 5 exams for each eye can be time consuming, is it necessary to define the pupil on *all* exams that are to be combined, or only on one (which will end up becoming the "most central exam")?

Yes the pupil does need to be accurately defined for each of the captured images. If the pupil is completely or even half wrong then it is generally easier just to redefine it (rather than just trying to move the existing points). 

Quote
Also, how closely does the redefined pupil have to align with the actual pupil (is it OK as long as it's concentric?); and does the software ignore obviously non-circular (mis-detected) pupils?

It is best to define the pupil as accurately as possible as it is used as part of the reference for recombining the images. While defining the pupil concentrically may work OK you will get more consistent results by having accurately defined pupils on each exam that is going to be used in the composite map.  At this point the software doesn't ignore non-circular pupils.
Senior Software Engineer
Medmont Pty. Ltd.

dmw69

  • Newbie
  • *
  • Posts: 25
    • View Profile
Re: Composite Eye - defining pupil
« Reply #3 on: September 24, 2013, 06:22:01 pm »
Thanks for the answer Michael. I had been moving the existing points, didn't realise you can just delete and make a new pupil - slightly easier.

Best of all would be if the software detected the pupil perfectly each time, but I suppose that's easy to say, harder to do.

Michael

  • Senior Software Engineer
  • Administrator
  • Hero Member
  • *****
  • Posts: 10149
    • View Profile
Re: Composite Eye - defining pupil
« Reply #4 on: September 25, 2013, 01:07:03 pm »
You are correct, it is harder to do ;)

Having said that now that composite captures are becoming more common we may need to revisit the pupil detect algorithm and see if it can be improved. However this will almost certainly be a project for a later date.
Senior Software Engineer
Medmont Pty. Ltd.