 All right. Good morning everybody. I think we're going to get started. Our first speaker this morning is Charlie Weber. He's one of our two excellent glaucoma fellows this year. He's been a real asset to us as residents and even admits his orthopedic injuries that he fractures elbow on the ski slopes this year. He's been ever present and always willing to help us as residents. So Charlie is going to talk about OCT and the assessment of glaucoma something that we're all very interested in. So here's Charlie. Thanks for that introduction Zach. All right. So as Zach said, I'll be talking about use of OCT and the assessment of glaucoma and this is mostly concentrating on RFL analysis and not into your segment OCT, which is a whole different topic. And especially at the VA where I'm working with residents most closely, the we get a lot of OCTs and I had a little trouble explaining to the residents why why is this important how can we follow patients and what are the important things to look for on the OCT scan and print out. So this will help to summarize some of those points and things to look out for to avoid pitfalls in OCT analysis. So just to review, there are two main commercial available types of OCT technology. The older technology is time domain OCT and the newer technology spectral domain OCT or a lot of times it's referred to as high definition OCT. The big difference is that with time domain OCT the scan depth is sampled at one point in time by using a scanning reference mirror and that produces the axial scan or a scan. With spectral domain OCT there's a stationary reference mirror and then the interferogram that's produced is then undergoes for your transform and then produces axial scans or a scan that a variety of different depth swish with one scan. So it's much faster, it can produce a higher resolution and it samples a larger area because it is faster you can sample more in a shorter amount of time. And in glaucoma analysis we're most concerned about the retinal nerve fiber layer in glaucoma disopteric neuropathy there's a chronic progressive loss of the retinal nerve fiber layer and detecting that nerve fiber layer loss is the kind of hallmark of glaucoma glaucoma damage and diagnosis management. So if there is a way to follow that nerve fiber layer loss that's really ideal. We do that a lot of times of course by just examining the optic nerve at the slit lamp using that as an assessment of the entire retinal nerve fiber layer of the entire retina and looking at that most closely at the optic disc. But there can be some inter-clinician variability and assessment of the disc and it's not a very standardized way but under an expert eye of course this can be an excellent way to manage glaucoma. So OCT tries to achieve a little bit more objectivity in those measurements and when an image is taken of the optic nerve head it produces on the print out a vertical a transverse and then an on-foss images. So here we have B scans and C scans and this is all a collection of the axial scans and for the optic nerve head analysis automated software algorithms will identify the end of Brux membrane and use that as the outline for the optic disc and then the outline of the RNFL is the optic. Once the the boundaries identified it produces an output for the clinician and parameters on this output are optic disc rim area, the total disc area, average cup to disc ratio, vertical cup to disc ratio and the cup volume and then comparison to the normative databases in addition to showing the on-foss image with the overlay of possibilities of damage and then the tomograms. But there can be obstacles in obtaining useful OCTs. They can range from acquisition issues so media opacity that can be from cornea lens vitreous and we'll see some examples of those coming up. Peripapillary atrophy myopic degeneration can cause problems in sampling or acquisition. Pupillary meiosis can inhibit scan quality and oftentimes patients will measure with a higher RNFL value if they're dilated versus having a myotic pupil even though scan quality may appear adequate with a small pupil. If the patient isn't able to hold fixation or gaze on the the target to allow the technician to complete the scan of course they can't really have an effective OCT taken and then if there's other optic disc or retinal abnormalities for example a coloboma that would impair the examination of the optic disc and nerve. And then there can be interpretation issues. How is the scan quality? Is there an image artifact? Is it reproducible from test to test? How does it compare to the normative database and is that a patient that you can compare to the normative database and what patients are even included in that normative database? And then does it correlate to the clinical appearance and we'll go through a lot of these. For example reproducibility there have been several studies that have shown that spectral domain is actually quite reproducible from test to test. Time domain not as much so it's a little bit harder to judge from examination to examination based on some larger studies. And then for example clinical correlation well does it fit with what you're seeing clinically? Is there asymmetry between the two eyes even though they might fall within the normal range for example? So when you look at an OCT there's three scan parameters that you should look at before interpreting the scan. One would be signal strength which is somewhere on the OCT printout. Centration of the image and then the alignment of the scan. And so let's go through some of the pitfalls or where these appear on the scan. So this is a lot of the examples I'll show are from the Cirrus OCT but the evaluate OCT holds true for any of the platforms. So signal strength here in the Cirrus is up at the top and you can see this scan in particular as an 8 out of 10 signal strength. So you need an adequate signal strength for an interpretable scan. Why is that important? Well you can't bill for a non-interpretable scan. So if it's non-interpretable it's just a waste of time and resources. It also measures the intensity and uniformity of the signal and so it looks at the overall scan. It's a scale of 1 to 10 typically and good is typically considered greater than or equal to 7. So if you have less than that you should view it with some suspicion or consider just repeating the scan. Next look for centration. So any centration errors can lead to RFL thickness defects centered circle around the optic nerve and then the optic nerve head in the center. Look at some examples of that. So here you can see the ANFAS image and there's a 6 by 6 millimeter cube that's outlined for data analysis and then within that a 3.46 diameter circle. So both of those should be well centered on the optic nerve. Errors in centration so here you can see in this ANFAS image that the cube looks well centered but the circle surrounding the optic nerve head is off-centered more to the temporal side of the disk and that can lead to errors. So here we're comparing in the normative database what's probably actually here to something that's a little bit more temporal and that can produce an artifact such as this one where we see temporal thinning but in reality if you look at the scan it's probably because it's somewhat de-centered. You can see that it does pick up some thinning there but again it doesn't allow you to compare effectively to the normative database. So looking at these finer points can help you at judging if these are quality scans. Scan alignment so if you have a poor scan alignment you may miss data points and cause a poor quality scan despite having a good signal strength. So it looks like a good scan just in a brief look you look at the signal strength looks plenty adequate but might have some artifacts. So here we can see a well aligned scan in the depth of the scan it falls right about in the middle. You can see that the software algorithm has been able to outline the RNFL in this case. You see solid lines throughout tracing the RNFL but if we have a misaligned scan in this case scan is actually too deep producing the scan to be too high in the scan computation and you can see that there's a break in where it's picking up the RNFL mostly because here at the most anterior portion it's lost its markings. And so if we look at the actual print out then it drops out at that area and you can see that it goes down to zero. You should never have zero on an OCT. If it's thin it will at least show something there. So this is just a problem and almost looks like maybe this patient had a little bit of pupillary meiosis and that was maybe one of the culprits not having an adequate scan in this particular patient. So is there movement artifacts? So if you look at Don Fosse's image here it's pretty obvious and see these lines that come across because the patient's eye was moving during a scan acquisition but sometimes this can be a little bit more subtle and so here it's obvious but if you look closely and see that there's a few lines of movement. This might even show up as a scan of adequate signal strength. What are other causes of partial loss of signal? Well here perhaps this is a quite opaque y-string from a PVD causing a loss of data and that section shows thinning on the nerve print out and that might show up as a shadow artifact here where you can see that the tracings of the RNFL are lost and all images lost wherever that shadows cast and that results in. So partial loss of signal can also occur in conditions like peripackie generations. So what do we do in these patients? How do we evaluate these patients with kind of anomalous optic nerves? It's a lot of patients that are referred to the glaucoma service. How do I evaluate this patient for glaucoma if I can't really look at their nerve effectively? So there's a couple of problems. One being that the scan acquisition is difficult and we'll show some more examples of that. We've already shown some things to look out for and then the comparison of normative database. So we'll actually look at who's included in those normative databases. So using an example of a patient with high myopia you can see photographs of the optic nerves here on the left of the screen and the OCT is really pretty much worthless in this case that it can't pick up anything for the left eye. Just can't pick up any landmarks. It looks like they had some problems acquiring the scan. Then here on the right we see multiple artifacts and it's having trouble accurately tracing the RNFL and other landmarks. So we look at normative databases. This was something that was a little bit surprising to me when I actually delved into this and looked at these a little bit closer. If you take Sirius for example which we've been looking at there's 284 subjects in the normative database that range from age of 19 to 84 and it was six US centers and one center in China to acquire patients for the normative database. Spectralis for example its only center was in Germany for its study and it had about 200 patients 201. So if you have a patient that's 90 they're not even represented in the normative database who would rely on those patients and then 80 to 84 range for that patient. If you have a patient using the spectralis that's of Asian descent this only used Caucasian patients and you might have to use some caution in evaluating those patients with spectralis for example. And then in terms of refraction in particular I'll draw your attention to this line so kind of the higher ends of hyperopia and myopia are not well represented in the normative database so that patient with high myopia if they have a refraction of minus 18 there there's no one really to compare to in the normative database for them and that's not necessarily even a parameter that's entered into the data acquisition process. So refraction isn't really taken into account might not be represented well in the normative database so there's our kind of our two problems acquiring a good scan of the optic nerve head and is the patient represented in the normative database. Well we can't do much about the normative database we have a patient with a minus 15 refraction minus 18 refraction but we can avoid the the optic nerve all together and use gaugling cell complex analysis. And so this is something that's becoming a little bit more utilized in clinical practice and what it does is it glaucoma is optic neuropathy will result in the death of retinal ganglion cells and so if we can sample the retinal ganglion cell layer we can also assess for glaucoma not just with the art of fellows mentioning earlier. And so the high resolution of the spectral domain OCT allows us to sample with that a level of image quality and accuracy. And so if we what the ganglion cell complex consists of is outlined here so we see that it includes the ganglion cell layer and the associated nerve fiber layer and inter plexiform layer and then the spectral domain OCT computes the ganglion cell complex thickness data from a seven millimeter square area of the scan centered over the phobia. And so here's what that looks like interpreted as a significance map so pretty similar to the the parameters before using the normative database this is an optic view scan. Green is average or normal and anything that's yellow is borderline and anything that's red is abnormal at the lowest ends of the spectrum. This could also be used to follow patients over time in theory just as we do with our retinal nerve fiber layer thickness. And so here you can see a patient in 2007 through 2010 we see progressive loss of the ganglion cell complexes time passes likely indicating progressive glaucoma. So when we evaluate an OCT there's really three scan parameters to review that we should use before interpreting the scan. Again signal strength, centration, and scan alignment. And so does anyone of the residents want to take a stab at just analyzing this? This is similar to what we would see over at the VA. Jim Bell you're at the VA right now I'll call you up. Unfortunately it isn't actually on this printout so that's not a very good example of that but oh sorry there it is. So 22 is an adequate for good yeah so I think we've avoided a lot of the pitfalls of earlier scan examples that we've shown. What would be the only other acquisition or interpretation issue that we would maybe be concerned about in this patient that I haven't given you much data kind of guess what I'm thinking. Right so just comparing to the normative databases and that of course looks like an adequate scan I'm sure it was for that particular patient but just realizing that there are some pitfalls for different imaging modalities and so they even give a warning on Heidelberg that it's a valid for Caucasian eyes only so if you're seeing anybody but a Caucasian you just aren't compared to the normative database I think it's probably still adequate for examining patients that are non Caucasian but there is just that kind of caveat or a warning that's right there on your printout. Is that functionally important and I see in the study that they show the differences between the different databases between one and two. I think it probably is not clinically that significant but it's just something to be aware of and I think where it's probably more important not necessarily race but some of those refractive area errors are evaluating those patients with anomalous discs. Yeah I don't think they do necessarily but the refractive area is probably more important because those are the patients that tend to have a more anomalous nerve. Yeah and that brings me to one of the final points when you're following these patients over time you have to look at each previous scan make sure that centration is similar from scan to scan whether scan quality adequate so if they have a scan of lower quality that results in a lower overall number for the that one out and look at the other ones in terms of progressive glaucoma so residents were at the VA looking back through when their patients are passed from resident to resident to even attending to attending and following this patients over time looking back through the scans making sure that each one is of adequate quality not just looking at the smaller numbers but can you compare to previous scans and is that an adequate comparison especially for example Zeiss forum and trying to create some of those progression analysis type software algorithms and it hasn't been validated to actually follow but I think you could presume that progressive thinning of the RNFL and OCT would indicate progression but ultimately I think it comes down more to clinician examination of the optic nerve and then progression on visual field test it's kind of the gold standard yeah exactly if you have a especially a spare stereo photo and that seems to be in our clinical practice what we're discovering that maybe in those pre-parametric glaucoma patients glaucoma patients that that can be some utility in OCT and detecting the presence of glaucoma in addition to the nerve valuation of the slant because in severe glaucoma if you get kind of almost complete wipe out the small changes that might actually show up on the visual field test at that point don't necessarily show up on OCT and where does it fit in our following of these patients over time they have a cup to disk of 0.8 and it looks like a nice healthy rim we send them for the OCT that comes back yeah that really reassures you and you've done your diligence to prove that potentially prove it and you the spectral domain OCT became commercially available in 2007 since pretty new technology we're still kind of figuring out how to use this tool most effectively yeah and retinas completely change Xana Park at UC Davis they're doing adaptive optics CT looking at individual okay thanks Charlie up next we have