Does it really have to be a fundus camera? Is it possible to use a
scanning laser ophthalmoscope (SLO) instead?

Although it is easier to build a fundus camera, the SLO gives a much
better contrast and it is easier to combine with other imaging
modalities, such as polarimetry. A couple of years ago, I built a
fundus camera in my OCT (optical coherence tomography) system for
retinal imaging, but if I would have to do it again, I would use an
SLO instead, because of the earlier mentioned advantages. However,
there are applications where a fundus camera gives better results.

Have you investigated the fundus cameras already on the market to see
if one could be adapted to your special need?  There are a number of
digital cameras using light available as well as scanning laser
imaging systems.   You may not need to re invent the wheel.

http://www.retinalcameras.net/

Dr Judy

 You may not want to or can't, but you might get more feedback if you
define (explain better?) the why of "special purpose".

Don W.

Please people, I really do know what I need.  Can anyone help with
literature suggestions?

www.richardfisher.com

I have been working with the Topcon instruments for a while, designing
digital camera upgrades to the film cameras originally used on them.

I think I have come to understand their essential optical principles.  
To my knowledge none of the many textbooks on optics and instrument
design treats this particular type of instrumentation, since it is so
specialized.  You must also have a theoretical model of the eye's optics
in mind when analyzing fundus imaging, although this is only lightly
treated in a very few textbooks.

The key element is a very high-powered, aspheric lens which more or less
matches the eye's focal system of about 80D.  The eye's focusing system
and the instrument's thus combine to make a 1X telescope, providing a
"window" into the eye for low-power microscopy with further optics.  The
asphericity is critical to compensate for at least two factors, one that
the eye's system is quite aberrant, and second the retina is a sharply
curved target.

Another aspect of the instrument is that it must retro-illuminate the
retina yet not through the center of the pupil where it would reflect
off the corneal surface and wash out the view.  So a hollow cone of
light converges onto the pupil, at a very steep angle that the eye
optics invert and spread broadly across the retina.  There is a
fundamental conflict between this retro-illumination and its strong
corneal reflection, versus the need to simultaneously image a naturally
non-reflective subject further on.  This is further complicated by the
eye's pupil being difficult to adequately dilate in many patients.

Practicalities intrude, such as the need to collimate in several
dimensions a large and heavy apparatus to the patient's jumpy eye.

You may also want to consider making a model eye.  I've used 80D
aspheric condenser lenses in front of an eye-sized shell coated inside
with a retinal photo.

Fundus examination is nothing new.

Have you looked at patents?

Colleges of Optometry (the state variety) have libraries which may be open
to the public.

This is true at least in NYC.

Also Medical Schools with an Ophthalmology Department should have libraries.

Tthere is surely one in Ann Arbor .

Even if not open to the public on a regular basis they should allow you
to sneak in once in a while if you explain your purpose.

Google Scholar with "fundus camera" yields 4000 answers mostly
to ophthalmology magazines which should accessible in the above
libraries at least the paper version of available.

Google Books with the same yields 400 answers.

Good Luck.