Optic Nerve Head Analysis
The development of the optical nerve head (ONH) segmentation algorithm is something I really enjoyed working on as it involved considerable inter-disciplinary participation to complete. Primarily this came from Dr. Douglas Anderson and Dr. Donald Budenz, both then at the Bascom Palmer Eye Institute. Dr. Anderson patiently tutored me on the anatomy of the optic nerve, the part of the eye where the ganglion cell axons exit and head toward the mid-brain, and what structural features could be seen with OCT. And Dr. Budenz steered the research toward clinical utility, assessing the efficacy of each parameter. Here is an initial study, a paper comparing the algorithm to expert readers:
Sharma A, Oakley JD, Schiffman JC, Budenz DL, Anderson DR.
This is an important contribution given that the standard structural measurement is still the stereo fundus photograph and OCT is, to many, an emerging technology. In many respects, this work helps bridge the gap between the use of fundus photography and the more explicit use of 3-d imaging. The following study, however, more directly investigates clinical utility:
Ability of cirrus HD-OCT optic nerve head parameters to discriminate normal from glaucomatous eyes
Mwanza JC, Oakley JD, Budenz DL, Anderson DR; Cirrus Optical Coherence Tomography Normative Database Study Group.
With any new measurement, the obvious question is to ask if it is reproducible? That is, is the algorithm accurate enough to detect real change, where change is perhaps the most reliable indicator of something pathological. This study indicates that, for their data set, this is indeed the case:
Mwanza JC, Chang RT, Budenz DL, Durbin MK, Gendy MG, Shi W, Feuer WJ.
The ONH algorithm, along with many others, is often discussed at meetings where expert clinicians communicate to colleagues the worth of particular imaging devices and their ability to make clinically meaningful measurements. OCT is, after all, a relatively new technology where standard measurements do not yet exist and perceived algorithm performance is often the only differential between competing devices.
Returning to the subject of this post, however, we present here a link to a talk that gives some background to this ONH algorithm. And here is another link where the algorithm was first introduced by Dr. Weinreb at AAO‘s annual meeting in 2009. Before Dr. Weinreb talks, Dr. Puliafito introduces the then new “Selective Pixel ProfilingTM” software, which I also created a little while ago…
