The Humphrey Field Analyzer has been one of the most commonly used diagnostic tests for glaucoma over many decades. For a better understanding of the diagnostic process we need to understand how the machine works. This will help us extract better compliance from the patient performing the test. The device, especially the software, is evolving, and has brought many new features making it relevant in today’s era. Dr Dikshit will be talking about what the machine does to the eye being tested, how it responds to the trigger, or the lack of it, how this impacts the result and how the device has evolved over time.
Lecturer: Dr. Siddharth Dikshit, DNB, FICO, L V Prasad Eye Institute (LVPEI), India
So, if you know this thinking man, we need to know not only how the man, the physician or the patient things we also know, need to know how the machine things and let me dive into the talk right away. I’ll be covering this topic under the heads of headings of steps of vision field testing differences between these strategies. So, you know we have something called Full threshold, SITA; SITA Fast and SITA Faster. SITA’s stands for Swedish Interactive Thresholding Algorithm and this is being the standard for quite a while as far as the Humphrey Field analyzer is concerned.
There are some common errors that can happen with any software and our person treating a glaucoma patients needs to be aware of these so that they’re not misled by these common errors. And there are some things new while some of these are actually knew some of these are old new because not many people talk about it, I’ll be going through this. We all know how a single field analysis print out of a 24-2 looks like, so this is what the print out looks like. Please do remember that I will not be discussing the print out or how to analyze the print out. But classically this has been divided into zones and if anyone wants to understand more about it they should go into an article by Ravi Thomas published in the Indian Journal of Ophthalmology. Every time I go through it, I definitely get something new from the article, so it’s an internal source of knowledge as far as visual fields are concerned.
I will go into what I’m going to talk about, and that is what does the machine do. So, the first step is determination of the Foveal Threshold. So, after the initial setting, the refraction settings, the patient’s data is fed into the machine. The first thing that will when the machine will interact with the patient is to determine the Foveal Threshold. Now this is what the ball or the center of the ball looks like. So, at the center you have this yellow light. I have made it black so that you can actually see it, but the light that you’ll see here on the machine is yellow in color. You have a large diamond which can act as a fixation point for people who can’t see small lights. But what happens during the Foveal Thresh holding is these four lights is where the patient focus, focuses. And in the center there’s a light blinking and that is these stimulus that the machine will provide. So, this is what is used for thresholding Foveal.
Now what happens during the thresh holding of the Foveal is quite classical manner or something that is quite well explained and understood but I will want to ask a question before we move ahead. So, this diagram is something that you might be familiar with and it’s a very popular diagram. So, if the initial stimulus here was 28 decibels the value of the next stimulus will be how much? Is it 24 decibels? 28 decibels? 26 or 30 decibels? Do you remember that this diagram is providing you a clue? And I will wait for a few seconds before you can answer and let me look at what answers come here. So, the initial stimulus is not seen so the machine goes through this process. Okay. So, if the initial stimulus was 28 decibels, what you can very clearly see from the diagram is that it will increase the stimulus or makes — give a stimulus that is brighter by four decibels.
Now, I’m not surprised to see that people have given various different answers because I was, as a resident, also confused here. Please do remember it in this way. DB means depressed by, DB is depressed by. So, a stimulus that is 50 decibels is depressed by 50 times the force whereas stimulus that is 40 decibels is depressed by only 40 times the force. So, you — the larger is the number in front of the decibels, more is the amount of depression or lesser is the number, the brighter is the stimulus. So, if the initial stimulus is not seen and it is 20 date decibels and the machine is supposed to throw a stimulus now that is four decibels brighter, it will have to reduce the level of depression by four decibels so it will show you a stimulus off 24 decibels.
Now let us see how this works in a patient who has glaucoma and for whom we are doing a HFA and he’s shown a stimulus for Foveal Threshold. So, typically stimulus intends to be shown in an area when the patient is not seen. You’ll keep seeing this graph again and again so I’ll just explain this once before we go into it again and again later. So, if the green area is the area which the patient can actually see, any stimulus that comes here is a stimulus that the patient is able to see whereas the red or maroon area is the area where the stimulus by cannot be seen by the patient. So, the aim of the machine usually is to show in the initial stimulus in an area where the stimulus will not be seen. So, let us say the initial stimulus is 28 decibels and the patient can’t see it.
So, the stimulus will run itself for about 0.2 seconds and then the machine wait for about 1.8 seconds for the person to respond. And if the person does not respond obviously because he cannot see it, it throws a stimulus that is four decibels brighter than the previous stimulus so, at 24 discipline stimulus is shown to the patient. If the patient sees it the machine knows that the actual threshold or the actual capacity of the desperator point to see a stimulus light somewhere between these two points. So, what it will do is, it will show a stimulus that is two decibels dimmer or income in with the stimulus at 24 decibel, the stimulus that will be shown next year’s 26 decibel. Now if the patient sees it, it will show a stimulus again which is two decibels dimmer. And now as this has landed into the non-seen aided, the patient can’t see it so we can kind of accurately or within the precision of plus minus two decibels say that 26 decibels is the pressure. Now do pay attention to the fact that the area or the line between the seeing and the non-seen zone has been crossed twice. So, this is called the double staircase method or the bracketing strategy.
Now what happens if the initial stimulus is already seen? So, a person with supermanized will see us 50 decibels stimulus also. But let us assume that this person has seen a stimulus of 28 decibels. So, what the machine will do, it will show a stimulus that is four decibels dimmer. So, a 32 decibels stimulus is not seen by the patient or the subject now and the machine knows that the stimulus, the threshold lies between the two points. So, it will show us stimulus that is two decibels brighter and what we will have is a 30 decibels stimulus seen by this patient here and 30 decibels becomes the Foveal Threshold because that’s the last stimulus seen.
So, do you remember that 50 decibels is dimmer than 40 decibels, 50 is the amount that these stimulus or the light is depressed by. So, you can imagine a filter being placed in front of a light off of 50 or 40 capacity, that will help it. And Foveal threshold no matter what the strategy is, he’s always did determined by the double staircase method or the bracketing strategy so that the area between seeing and not, non-seen areas is cross twice.
Now what will the machine do next? Once the Foveal threshold is determined you can kind of guess that this is the stimulus that is always going to be seen by the patient and depending on the location. Farther is the point the threshold of that point becomes a little lower than the Foveal, so that is kept in the mind and the machine now understands what needs to be done.
So, the first-four points which are tested at the four cardinal points or the primary points. So, these are the points which had located on a 24-2 or 30-2 around nine degrees from the center. So, these four points have located around three degrees from the center and the next step, the distance separating the these four points in the next row in 6 degrees. So, these four points are 90 degrees from the center and these are tested twice.
So, if you see a natural print out of a full threshold test, you will see a number written without the parenthesis or the brackets and the number will be written within the brackets. So, these two numbers indicate the values of the threshold that this these points have on being tested twice. And you see here that the patient is an extremely well by seeing the same stimulus on both occasions. So, this gives a very accurate description or thresh holding of these two points.
Now in addition to these four points in full threshold program, there are six more locations which are randomly selected. So, if you can see the purple circles devolved circles here did not points which have been tested twice and have the values written twice once with the brackets at once without the brackets. So, these are the points that have been tested twice. Now why does the machine do it, they calculate these on the basis of that. They calculate the acceptance interval. So, the machine now knows based on these points how what is the average response time of this patient.
So, a patient who is old may take a little longer than a younger patient; a patient with advanced glaucoma may take a little longer than a normal or an early glaucoma patient to respond to the stimulus. So, each person has the right to determine their own average response time and then the machine adds around 0.85 seconds to it so that if there are certain points which are more disease the patient still gets enough time to respond to these. It also calculates the short term fluctuation by looking at the difference between the two values when these points were tested twice. Beyond this the test proceeds with the double case, double staircase or the bracketing strategy in the full threshold from strategy.
In SITA, this does not happen. So, SITA does not test the additional six points. So, how does it get this information, it uses complicated statistical functions and if you are in an exam, the examiner will be really impressed to hear the term the Bayesian posterior probability function. So, this is the exact methodology in weight by which the machine determines the expected threshold of each of the points based on the four cardinal points the age of the patient. And if the patient has previous visual fields, it can do a modeling. And then what it does is it calculates the information index continuously. So, once you have a point where the machine things that threshold should be 24 decibels, you know that it cannot be supersized or say machine because it’s working in decimals also say 23.7.
So, the machine will accept a response which is, say, plus minus one decibels, so one decibel is not important. What is important to understand is that there is a confidence interval against this mean within which it will accept the result. So, if somebody sees a stimulus at 24 decibel when the estimated the threshold is 23.7 and if 24 decibels is within the confidence interval, the machine will accept it as a stimulus. So, this information index keeps updating continuously so does the visual field modeling. So, even if it does not test the additional six points because the machine is taking all the other points into consideration it generates a very robust model or estimate or the estimated threshold of each point.
And then when the test is over the machine spend some time in analyzing the uh responses that the patient has shown and then adjust the estimates accordingly. And the reaction time also depending on the patient as you saw in full threshold initially the time, the response time is calculated in the same way. The longest response plus point 0.85 seconds but then it continuously keeps on existing the same depending on the patient whether the patient is responding fast or slow.
Now after these four or 10 points attested, what does the machine do, then the business begins. So, till this time it was preparing and then it starts thresholding the various retinal points within the area that is being tested. So, if you look at it threshold, what one must understand is that threshold at a given point is the intensity of stimulus that has seen 50% of the types. So, even when the threshold is, say, 29 decibels at this point when that 29 decibel light or stimulus is shown to the patient 10 times, he’s likely to miss it five times. So, one must reassure the patient that you are not supposed to see 50% of the lights and so you should not be anxious or try to change the lights because the machine will try to go into a point where you don’t see the light 50% of the times very much like a traditional viva or vice versa that happens when the examiner didn’t seem to and takes you to a point where you can’t answer questions. So, this is where the point will be and that is what is called the threshold.
Now, how does the machine do it? Now, this is different for different strategies. Now,
we have already discussed the full threshold algorithm and it is pretty much the same bright stimulus that is not seen is made brighter. And then you have a depression by two decibels and the threshold is or the area between the seen and not seen area is cross twice, and the last seen stimulus is recorded as the threshold for the point. However this does not happen in SITA. How does SITA do it differently is by calculating and acceptance interval. So, the machine for this particular point already has decided that the stimulus is going to light between, the threshold is going to light within this blue rectangle. And if the person sees a stimulus within this area, I will accept that as the threshold.
This is calculated on the basis of age, the data from the cardinal points and is continuously updated using the adjoining points also. So, this is what the SITA USB is this calculation. So, you see that the initial stimulus was 28 decibels then of four decibel brighter stimulus is shown with the patient sees but the machine is not happy enough with this, it wants the stimulus within this blue rectangle. And so when the next stimulus around 26 decibels has shown and the patient sees it machines is okay happy. This person is doing exactly, this rectal point is doing exactly as I expected and I will accept this response as the threshold for this point. Do remember here that the line between the seeing and non-seeing area is crossed only once. It is also possible that when the first stimulus is shown and not seen, the second stimulus can be seen directly by the patient and this will also be accepted by the patient without going down the ladder.
So, this is pretty much how the machine does it but in case a point is severely depressant does not do well the SITA will also give an opportunity by the double case, double staircase of acting strategy for exact determination of the threshold of that point. That is not routine but that opportunity will be given to a point that is not working well.
Now SITA Fast is pretty much the same as SITA Standard, the only difference is that narrow zone of SITA Fast, SITA Standard is much wider Bandy Zone that is chosen for SITA Standard. So, the estimate is not that accurate because of the virus wider acceptance window. And then as soon as the patient sees a stimulus within this zone, the machine says okay, I’m happy with you, let us go ahead.
So, do you remember that all points at threshold using the double staircase of the acting strategy. SITA Standard does not use double staircase except when the points are really depressed. SITA Fast works in a similar manner but with a wider or a greater acceptance window with greater confidence limit.
Now when this is happening there is something else also happening, I don’t know if it’s correct to call it Step 3 because this is happening simultaneously. So, this is what the reliability parameters are and I’ll be only talk — excuse me. I’ll be only talking about the false positive and negative errors because this is different between the machines on and the programs. And this is something that is not very well understood. So, I realized that when I was talking to my residents and fellows and the problem is that is twofold. One thing is that this is a proprietary software and very little is actually often to open related to the software. And the greater problem is that many standard textbooks actually don’t go into this. You need to read a book on visual fields or standard glaucoma text book in order to understand this, but I’ll bring it to you straight away.
So, false positive errors in full threshold and Fastpac is determined by throwing a fixed number of stimuli in a, which are very bright. So, a three point which has already been threshold, it’s here 25 decibel, it will show a stimulus that is probably even 7 to 9 decibels brighter than to stimulate. And if the patient does not see it, it is not the retinal function that is making them incapable of seeing it but the fact that the patient is not able to see it.
So, how does SITA do it? The SITA does it in a different way. So as soon as a stimulus has shown, the SITA has in mind acceptance window for the duration. So, it knows that the stimulus has to come within 180 to 200 milliseconds. And this is the green zone when it will accept the responses through positives. What happens if it comes beyond this? Then probably the patient is not responding to the stimulus and that is considered as false positive. Also if it comes too fast while the stimulus is being shown, then that also comes sense false positive. So, that is the manner in which patient is shown diagnosed as false positive in SITA.
I’ll just go back a slide I think I mentioned something wrong in the previous slide or I’ll come back to it later. So, false negatives in SITA as well as full threshold is determined by showing a very bright stimulus brighter than the thresholded previously thresholded point. And any stimulus that is not seen by the patient is actually the false negatives. So, what happens in false positive is that the stimulus is not shown but the sound is produced, so that that is where I was wrong about the full threshold and false positive. So, in false positive tests the machine it seems to be poised and it’s time for a stimulus much introduces the sound but does not show a stimulus. And if the patient response to that, it means that the patient has not responded actually the light but just to the feeling of or the urge to see a light and that then the patient has responded, that is false negative, false negative, false positive. False negative is when the patient or the subject actually fails to see a similar that should have been seen by this patient.
Now the real job happens after all this has been done. Now this is something that a lot of people can do any machine of worth can do but the USB of these gold standard machines more than just some realizing what you do after this. So, once you have the thresholds, you can calculate the total deviation. A totally deviation is nothing but point by point difference of these thresholds from the expected value for age matched for age in normal individuals. So, a normal person, so if you see here if the threshold for this purpose subject here is 27 decibels and you see minus five written here, it means that this point in normal individuals has a threshold of around 32 decibels. And this is five decibels below the expected value and that is why you see this minus five here. Minus one means it is one decibel lower than the expected value and then you generate the total division probability plot. And if you see there are four categories of P values that can be marked here on a 24-2 and 30-2 and the denser and darker it becomes the more is the chance of this being glaucoma. Now I’ll come to that a little later.
What is very interesting is how the pattern deviation plot is calculated and this is something I found a little difficult to understand earlier and I tried to make it easier for you. So, you see a total division plot here on the left side and in the right side of the screen every point has been corrected by some number and you probably are seeing every number defect depth or renovation from normal has become better. So, three has become zero, two has become minus one because it seems like there has been a correction of around three decibels at each point. Because these corrections happen in decimals and they’re rounded off by integer rounded off to the nearest integer values. You can see some difference of around correction of around 2 to 4 decibels, and that can happen because of the decimal based calculation and rounding off.
I will give you a question here before I will answer it. True about generation of pattern deviation numerical plot from the total deviation numerical plot is all except. 85 per — 85th percentile is the cut off for generalized depression; the value of the seventh highest point is directed from the defect depth of each point in tenders to the values presented or rounded off to the nearest integer. I have given the answers of this one of these and it is not calculated for the fields performed with stimulus size. I’ll just wait for 30 seconds and take this break to drink some water. Okay, a lot of you have got it right. So, what is wrong is that the value of the seventh highest point is directed from the defect depth of each point in 10 days to program. Now when I explain it to you in detail you will understand this.
So, when the pattern deviation plot is being calculated or pattern deviation values are being calculated even on a 30-2, it is done only on points which are included in 24-2. So, all the peripheral points except the two nasal points are removed. Also removed are the three points exist into the physiological blind spot. So, these are the three points which are removing along with the points which are excluded from 24-2, so the calculation is done identically for the 24-2 and 30-2 program.
Now, what I have done here is I have noted the defect that for all these points here on this Excel sheet, and you can see that we have the threshold point, points for thresholds for all the 54 points line here. And now what the machine does next is to remove the three points which are existing to the physiological points so the values of those three points are removed. And at the end what we will have is only 51 points which will be used to calculate the patterns and the deviation. So the minus 17, two decibels and four decibels is something that is removed from it. And what it leaves with you leaves you with is 51 points.
Now what the machine does is arranges these in an order from the smallest to the largest number. At this point this, this column of 51 points can be represented in a graph like this. So, you see the smallest number here and the best number here. What the machine does now is it calculates the 85th percentile of the total number of points which on a 24-2 and the 30-2 is seven points. So, it removes the first six points and threshold of the seventh best point is taken and that is adjusted assuming that there is a generalized depression and we need to we will be localized depression. So, this three will become zero and three is the number or the machine actually goes into decimals but the integer value is three. All the points, from all the points this number three will be deducted. So, you see three has become zero, one has become minus three probably because of some integer differences but almost all the points would be corrected by something between two to four decibels. And that is how the pattern deviation plot is calculated.
Remember that total deviation is the deviation from the expected values of the age match normal individuals. Pattern deviation plots removes the generalized and assumed the 85th percentile location and the highlight, this highlights the glaucoma related field loss. Now the 85th percentile for attenders to which test 72. [distorted] [00:30:33] this will not be the seventh best point, it will be something different. It will be a larger number probably 9th or 10th but it is not going to be the seventh, the seventh highest point is considered for a 24-2 and the 30-2. The number of the location of this point will be different for then there’s two points.
Now, we’ll come to the Global Indices. Now this is what is the numerical representation of the entire visual field is, and a lot of calculation goes behind the scenes in this. Mean deviation, pattern deviation has been around for quite long and the mean division plot or the mean deviation value is calculated from the total division plot. And the pattern Standard deviation value is calculated from the pattern division plot. These have a decibel value and the P value attached to it.
Now in this table that you see, what is true, there is a P value attached to the pattern standard deviation. What does it mean? It is 99.5% likely that this is glaucoma. It is 95% likely that this is glaucoma. This is an early glaucoma with 9% vision loss and up to 5% of normal subjects can have similar patterns standard deviation, what is true out of these. Okay, great. I am very happy to see that most the majority has got it right because what it means is that around 0.5% of normal individuals at this particular point we’ll have or will have a mean deviation like this. So, what it means is that it does not mean that the rest of 99.5% have glaucoma but it means that 0.5% of normal will have a depression like this.
Now it is our job to figure out what has caused this depression because even cataract, ephedra that a little degeneration can also cause it, it does not mean that this is glaucoma. At the same time when you see the pattern standard deviation it means that 5% of people can have a pattern standard deviation like this. So one in 20 people so a room classroom where they’re 40 people sitting two of them, two of the brilliant doctors are going to have a pattern standard deviation like this. Now a patient who is doing as well as two doctors cannot be shortly you cannot say just looking at these numbers that they have glaucoma. You have to look at the larger fiction you have to look at the pattern of the glaucoma that is there, you have to correlate clinically.
Similarly, 90%, 91% VFI does not mean that it is glaucoma. It can be various causes including the posterior capsule cataract and this has been macular degeneration or even a hemorrhage on the Foveal. So, do you remember that these patterns are not affected only by glaucoma, these numbers, these patterns are not affected only by glaucoma and multitude of diseases can do it. And it is our job to find, and these can only be suggestive of the abnormality and not off glaucoma. It is our job to figure out what has caused it.
So, the VFI is a very intelligent index. And the reason to generate this was that many times the central points which are really important for vision we’re not even if they were involved, they were not really showing up in the terms of mean deviation which took an average of all the points. So, the VFI gives a lot of importance to the central points. And the second thing was that the mean deviation gets affected very significantly by the media opacities like cataract and reflective errors are a patient having a poor pair of glasses are one who’s being tested with a lens that is not clean. So, for this reason, VFI was developed and it largely compensates for these.
One of these situations where the VFI does not work really well is a posterior sub capsule cataract. So, one has to keep that in mind. So, the mean deviation is represents the generalized depression. The pattern deviation is likely, less likely to be affected by cataract and other generalized depression caused by cat alight and similar technologies. And even a feed of less than 5% 1 in 20 normal individuals will show similar depressed levels, it does not necessarily represent a 95% chance of glaucoma. But it means that this number is present in 5% of normal people so all people also. So, one must keep in mind that until issue prove the disease, do not consider a percent degree disease just on the basis of these numbers.
Now, there’s some situations which are tricky and I thought it will be good to discuss these. Now, this is a visual field a 24-2 print out in front of you. You can see the gray scale, the full threshold, the total deviation plot and the pattern deviation plot. Now, you will expect the pattern deviation plot to improve and show less defects than the total deviations plot. What you see here is a result of a) learning effect, weber’s phenomenon, advance glaucoma or just the depression has switched off and the fields are unreliable. Which of these has caused the problem here? Do you look at the full threshold values, the threshold values of points and that will give you a good idea of what is causing this.
Wow, [Distorted] [00:37:12] speak that if you have actually chosen the right answer which is advanced glaucoma. Now why this is happening here is if you search for the seventh best point, the seventh best point is probably something like six or seven decibels. So, the machine is misled into believing that almost all the depression that you see here of around say probably the number will be 33 or 31 decibel is generalized. No, it is not generalized depression because of media capacity but it is extensive glaucoma causing this generalized depression. And the total deviation plot here is more representative of the disease than the pattern deviation plot. So, these are the worst affected points which still show up here but the fact remains that almost all of these are glaucoma. So, and over correction like this used to happen earlier but does not happen here, does not happen in the newer machine. I will come to that after the next question.
Now what you see here is an overview print out of 24-2 of the left tied of a particular patient. Can you please choose an option as to what is the explanation of this VFI falling from 60 decibels, 60% to 40%? Is the fields unreliable or does it means that doesn’t need for urgent surgery? Does it mean there is progression or none of these? Okay, 63% of you have said it is progression. It is not progression. It is not progression and this is one thing that in HFA machines we have to be very guarded against. You see, what happens when the patient repeats the field test? You see an alarming progression from the VFI falling from 60% to 40%, or was it 80? No, this was 60%. But when the patient repeats it, what happens here is there’s an improvement and the VFI even better than the first one. Now, why this has happened, this is very interesting.
What happens here when you have a pattern deviation plot generated, the machine generates the value of the VFI the pattern deviation plot and the pattern center deviation value which is much better than the mean deviation value. But as soon as there is even a much production of two decibels which is much within the test-retest value of the HFA machine, the total deviation plot is not generated because the machine knows now that in advance glaucoma, if you generate a pattern deviation plot it is going to be much better than the total deviation as it was happening on the top, top level. So, you see — oh sorry, it’s going to be much become much, much better than the actual diseases because almost all the points of depressed and then correct for corrective and the most depressed points because of glaucoma.
However, then the person repeats the test. He goes back to mean deviation that is very similar to the first test and the machine all generates the pattern deviation plot again and the VFI is calculated from the pattern deviation plot. So, the clue to prevent yourself from being misled by this is one. You will see that the mean deviation is still within 1 to 2 decibels. So, mean deviation within being within 1 to 2 decibels of the original test and VFI dropping by 20% is not explained by the disease. There’s something else that is going on.
Also, if you look at the grayscale, it’s pretty much the same. The total deviation plot is pretty much the same. And the biggest clue that you will have is that the pattern deviation plot for this particular test will not be generated. So, you should just repeat the test once more. And if you still doubt that there is progression, you should look at 10-2 for detecting the progression here. So, one has to be careful about this. What is a little funny is though on 24-2 and 10-2, the machine will not generate a plot, a pattern deviation plot because the field is globally depressed. It still does it for 10-2. And what you see here that in advanced glaucoma with severe depression on the total deviation plot leads to a completely normal pattern deviation.
So, though our tendency is to look at the pattern deviation plot right away and while somebody who’s starting looks at the grayscale, the person looking at the greyscale will be more right valid here. So, if you put this into the context of the thresholds, it defect depth and the grayscale you know that this pattern deviation plot is falsely normal and it probably means that almost all points there are depressed to a very significant extent.
Something more about 10-2, if you look at this image you will see the 24-2 plot and the 10-2 plot of both other patients, of the same patient of the right eye, what does this mean? Does it mean that optic neuritis can mimic glaucoma? Does it mean that 10-2 may detect more severe disease earlier than 24-2? Or does it mean that grayscale should not be, should be used to decide whether 10-2 is needed or not? And, or simply a fact that the 24-2 and 10-2 do not correlate here and there’s something wrong with the machine again?
I like the questions because it gives me a break to drink water. Okay, great. Almost all of you have got it right. And you are right in thinking that the 10-2 may detect more severe disease earlier. Why does this happen? So, if you look at the 30-2 and 24-2, they have the same principle of basically, this are points being separated from each other by six degrees and this central points separated by each other by three degrees. However, when you look 10-2, it not only limits the area to the central 10 degrees represented by this circle but it has many more points within this, within this 10 degrees.
So, if you look at these central 10 degrees, I will use these points. So, the orange and the green points are the points which are tested by 24-2 and this gives you a count of 12 points. Whereas, within the same area, the 10-2, tests 72 points — sorry, 68 points and that gives you a much denser appreciation of how this glaucoma. So, if a glaucoma looks like this, it involves only one point on 24-2. It may actually be covering the entire area of 10-2 without affecting any of the other points but only one – the point on 24-2. And that’s exactly what is happening in this field that you saw previously in the question.
So, in this question I’ll just go back to the slide. Only one point was getting affected. But in 10-2, you see everything is getting affected and that is because of the density of the points that are being tested. And the example there what is happening is only this point is getting affected whereas this look actual location of this glaucoma is the entire 10-2 fields here which is looking in a much denser manner in the central 10 degrees. What is also important is traditionally, we have asked for a 10-2 when the central points have been affected before central points are 24-2. But one has to remember that even if these paracentral points are actually picking up a defect on 24-2, it may be a much largest glaucoma in 10-2. So, a 10-2 is indicated if any of the central or the paracentral points are affected and then many publications now where these central 10-2 has shown a large glaucoma, despite the central point not being affected.
What is also true, so if you see here that the central points are not affected on 24-2 but 10-2 shows a much latest glaucoma. And what it does is it also picks up progression faster than 24-2 because it is looking at points much more closely than the 24-2. So, 10-2 is indicated whenever there is central or paracentral points that are involved in 24-2, and it may detect defects as well as progression earlier if this glaucoma are central.
Now, this has been a big criticism for Humphrey and they have tried to address it. But before we move on, can you please identify this print out? What is this print out? What does it look like? It is 24-2, yes of that there is no doubt. Is it SITA standard, SITA fasters, full threshold or something else? This is pretty much a straight forward question. I expect everyone to get it right.
Okay. Now it looks like not everyone has got it right. Okay, let me go back to the presentation. This is 24-2c SITA Faster. Now you can probably see that this is SITA faster when you look at this strategy, and you know that this is 24 degrees being tested because of the pattern of the points. Now, why is this not 24-2 SITA Faster but 24-2c is because you see these numbers, these are not normal locations that are tested in a 24-2. This is some, there’s something wrong here. So, what one can easily say that there is something different and the only option that is different here is 24-2 SITA Faster. SITA Faster, it is because obviously it’s written there and it is 24-2c. And this is the new member of the family which has addresses a large number of questions that were raised earlier.
So, what it does is that study because studies have shown that there are a lot of points on 24-2 between these points which are common areas of location, and so, that it has added these locations in addition to the traditional 24-2 points. These 10 points have been added and they’re much more likely to pick up scotomas which were hidden between the points earlier. And that seems to be the case and we have the newer strategy which takes even lesser time then the SITA program, when the testing time is concerned then the SITA standard. And this it does by many ways, so it starts with a much brighter much similar which is much more likely to be the closer to the actual threshold and does not always intend to cross the seeing and non-seeing areas. It intends to throw a stimulus right in the area of the act – acceptance window. And it takes care that primitively blind spots are retested. So, the number of cardinal points tested twice zero. So, there’s no point that is tested twice. But a robust modeling makes sure that you have good response acceptance window and a response interval, acceptance interval product thresholds.
So, without compromising on the quality of the field test, it actually does the test much faster. It can be used for glaucoma screening definitely, there is more and more evidence coming in favor of this being used for glaucoma patients. But that is not the current practice and we probably will need more evidence on that before that becomes the routine.
I will at the end bring something that is very interesting. If you look at this print out, this is a very interesting colorful print out and we are out of the zone of the black and white dull looking print outs of the Humphrey Field Analyzer into this colorful field plus OCT. So, what it does is it looks like the OCT print out, and if the inferior and nasal area both abnormal in OCT, it’s superimposes that area of the OCP is reversed. Inferior becomes superior, and superior becomes inferior. And it maps the color coding on the visual field. So, you can know whether the OCT and HFA represent this same abnormal areas. And it is very well correlating here. You see the localization of the defect on visual fields is pretty much the same as the localization of the fields depressed points on the OCT. So, thin points OCT and the depressed points on HFA are correlating here in the right eye. Of course, the OCT shows a much larger area of involvement which is expected. But then none of the points here that that depress on HFA that are beyond the OCT. So, this is very well correlating print out. So, this is a combined structure function report.
What is great now is not only this but it also gives you a Combined Structure-Function GP or the Guided Progression Analysis. So, you can look at the visual field index, okay, it’s going down a little bit. You’re not very sure if this is glaucoma. Mean deviation, this is also going down a little bit. But what is very reassuring is that the average ganglion cell layer thickness on the macular program or the GCL analysis is not declined. Neither has the RNFL significantly over time though all these combined probably mean that you need to be a little more careful because all these graphs are pointing towards the downward slope. And there is no harm in this patient when three out of four are pointing towards a downward slope to augment the treatment.
So, when you look at this print out it gives you a summary of five years of OCT or five visits of OCT RNFL and the GCA and the visual fields. And this is great for the glaucoma specialists and care of the patients that you have in your clinic. At the end, we know that the more Robust techniques take more time so we need to customize. Patient who are very old are unable to sit for a test for 10 minutes, when you do full threshold SITA fast maybe the option to go forward. SITA Fast, SITA Standard and full thresholds now can be mixed with each other to generate the GPA which is great. Glaucoma, the initial tests should be 24-2, if you are screaming it can be 24-2c. If the central or paracentral points are involved, please do a 10-2 before calling it a pre-parametric lost.
The SITA Standard and full thresholds differ in thresholding and false positive determination. And the pattern deviation plot better than the Total Deviation plot is suspicious, watch especially if the mean deviation is close to 20-, minus 20 decibels and you don’t see a pattern deviation plot generated and the Combined Structure-Function GPA is an excellent decision making tool to look at the overall progression of the patient.
Thank you so much for your attention. I will now try to go into the question and answer. You can put in your questions here, I will try to stop share. Okay now, I can see this.
So, Mohan Raj is asking, if we skip the Foveal, then at what threshold point the machine will continue? It will give a point which is very bright. So, instead of starting close to the expected threshold, it will start at a much brighter stimulus or much dimmer stimulus and take a little more time to threshold, reach the accurate threshold, so that the testing time will increase.
Okay, so but not performing a performing in Foveal threshold like the end is going to increase the testing time, testing time and not kind of help? So, the Foveal Threshold when you examine it is just going to add in a kind of estimating the threshold of a point makes the test faster.
How would the cardinal points are measured in SITA? The cardinal points of the four primary points are measured by the double staircase or full thresholding program. In the full threshold, there are additional six points that I tested and this doesn’t happen. The additional six punch don’t happen in SITA Standard.
What is meant by Fovea off or on? So, very, very interesting. Why would someone tend the Fovea off? Now someone who has a macular hole or any RMB or a Central PSE may not do very well in when they’re testing the Fovea. So, there it may be useful to not look at the Foveal Threshold at all and go directly into the glaucoma test which looks for not the central point but the central, paracentral points and beyond. And then it kind of, it takes off there. So, if you have a central, a disease that affects the Fovea, turn Fovea off, let the machine run by the testing the other points. I will not answer the question on near vision glasses beyond the scope of this discussion. And well, this will described in textbooks.
Okay. Now, why 85th percentile is chosen? Now I don’t know. I [indiscernible] [00:57:18] excellent questions and I don’t know why 85th percentile is chosen and there is no data said that the authors have published why 85th percentile is chosen and not 75th or 80th? I would believe, and so people who have taught me believe that it’s slightly random choice to choose the 85th percentile. And well, it’s slightly random probably but probably, in this field and [indiscernible] [00:57:51] may be the best person to answer these questions.
What if the patient is in low vision category? I think then you see the minimum vision activity, you can do a visual field test even for a patient without 20 by 200 visions. We have seen multiple patients do it. But what happens then is that you start getting a lot off unreliable fields. You start getting a lot of test, retest variability. Now, is that field going to be useful for you when you’re trying to diagnose glaucoma? If the patient didn’t have low – because of glaucoma there is no doubt in the diagnosis of glaucoma, it’s not going to help in diagnosis of glaucoma. If you’re looking to look for progression then it is not going to help because there is already going to be a huge test-retest variability.
So, even if I don’t have correct answer what one can do is to do the tested two or three times and see if the patient can produce a consistent test. If the patient does the test consistently with thresholds of each point, when you look at the three printouts or two printouts that you have are very similar then you can keep on doing the visual field test for this patient with no vision. Even if it just 20 by 320. Some patients do it well even then. Some patients that went 20 by 100 can’t do it well. So, you have to see whether the patient has the capability of producing a consistent result. Otherwise, you have to look at the general person advanced patients. They have very good perception off their vision. Every parameter looks good and the patient tells you, I — they can’t see things well as well as I see used to see it earlier. You have to take that seriously.
I will probably move on to the next question. Decibel can be used for sound as well as light. Okay, you want me to explain which [indiscernible] [01:00:10]. So, the difference is exactly what is the difference between average and weighted average? So, if, say, somebody has given six tests in a near during while they’re doing of course, you can say that we — I will take the average score of the six tests and give each this equal weighted and your average score is this. Whereas, then I, I can also say that the test that I have done at the end of the year is really important and it was the much larger of that test, I can say that I will determine 50% of the total 100 marks on the basis of that one test and the rest of the five test that averages going to consist or constitute or give me the rest of the 50% marks. So, that is the vintage that you give. And the VFI, what it does is gives much more weight is to be much more important central point, slightly less important to the paracentral point, even lesser importance to the point beyond that. So, it is a weighted average and the score rate is calculated, keeping in mind that some points are more important for the preservation of vision. Yes, stimulus five doesn’t have a normative data base or because it is something that is going to have a huge amount of test-retest variation.
Weber’s phenomenon, please look up the book on visual field in a dense cataract. Okay, pattern deviation plot, so in a nuclear sclerosis, the pattern deviation and especially VFI are not affected that much. In a posterior subcapsular cataract it does get affected much more.
Grayscale, okay, I’ll just sum it up in in a few seconds. So, you have four levels of depression that comes in the pattern deviation probability plot, and it looks at a very narrow range off depression. So, it looks at possible range of depression from 95% depression to, say, a depression of 0.5%. So, it looking at a very narrow range and it is divided into four possibilities whereas the grayscale covers the range from 0 to 50 decibels, a very large range. And it also gives you eight sub classes in it. So, what is the advantage of grayscale is any change that is not happening within this 5% of people or even the lowest, the most depress ranges will still be pick up by grayscale.
So, while it is picking up more, it is much more likely to be affected by things like pupil dilatation cataract, refractive errors, so that limits is used. But it can sometimes pick up changes faster than the pattern deviation probability plots and that is where it is used intelligently by glaucoma specialists. Otherwise, the sole purpose is to look at the pattern of this glaucoma. Do not try to use it beyond that, use beyond that has to be done very carefully. But in some patients where the changes happening in points which have a good threshold that can show up in grayscale earlier than the back in deviation plot.
Okay, Ravi Thomas, just type Ravi Thomas automated static perimetry Indian Journal of Ophthalmology. The book to go for is Automated Static Paramilitary by the founders, okay. I’ll skip a few questions, 30-2 does not need to be using glaucoma at all. You do a 24-2 but if you see most points that the depress and only central and paracentral points have anything. So, if the four or 6/7 points are the points that you’re looking at, there is no need to do a 24-2 at all because those will all be covered in 10-2. But if you see points beyond the central 10 degrees that have any threshold, you need to take care of those points also. You cannot neglect those points because progression can happen even beyond the central 10 degrees.
Okay, GPS still not available. I will try to get this message to the makers. GPA, I would love, but you actually have — we a plot mean deviation based plot now for GPA. And that is where the — it can kind of substitute the GPA plot that you have for 24-2. I did not cover GPA because it would come probably under progression and it would have been too much for the talk here. But you have, with the mean deviation based plot, now generated for 10-2 in newer software. You don’t have to change the machine, just update the software.
Okay, anything less than 16 years old phase you don’t have normative database than many more pediatric perimeters coming up, hopefully they will do the job. 24-2 you want to do because a 10-2 tests a lot more points it takes much longer. 24-2 is the balance between testing the central points and the time taken for the test. So, that is the advantage that 24-2 is going to provide over 10-2. It will not replace 10-2. Never, never mistake being for suggesting that 24-2 can replace 10-2, but the first test to cash those central point involvement or let me put it in a way that it is better than 24-2 in picking up the central points. It is definitely not better than 10-2. And if any of those central are involved, you have to do 10-2, very good question anonymous certainty. No, there are no classifications as far as VFI is concerned.
Okay, one very, very good question. How to avoid that unwanted test error? Look, the first thing is that the patient should be reassured. As I said again and I can’t stress this enough that you have to tell the patient that you’re not supposed to say 50% of the light. And if the patient keeps on chasing, the test becomes more and more unreliable. The second thing is that you have to tell the patient that you can take this remote looking, remote like things and you can pause the test. So, if you keep the button pressed, it will pauses it. So, you can pause the tests, take a break, have a coffee, you can’t obviously have a coffee there but a break to rest your eyes and then the moment you release it in a few seconds, the test will shut again.
Okay, the model it’s I believe available in even the two eye, you have to update the software. It is definitely available in HFA three and compatible OCT I am not sure which OCT are compatible. But even the 5/7 year OCT that we have with us, these Zeiss one gives a combined desire. It has to be the same company from Zeiss and probably would be the same for other companies if they have this combined report.
You have visual function questioner abnormalities and 10-2. Okay, this is great question. And what is very important to realize that the 10-2 being closer to fixation does two things. It affect the vision much more, much more severely than the peripheral effects, because when you’re looking at a face, you’re looking at the 10 degrees. Imagine you’re looking into someone’s eyes and not being able to see the mouth how difficult it is to communicate with someone when the mouth is covered. Does it look normal that you cannot see the mouth but you can see only the eye when I’m talking? It does not, right? So, that is the importance of defects on 10-2 fields and that does affect the visual function more than the 24-2. Mesopic conditions probably not because it’s the peripheral retina that works more in the mesopic conditions than the central retina.
Okay. I will skip the questions on interpretation of visual fields. I will not — probably, it’s a different talk altogether. I’m sorry, I would have loved to talk about those. Which is, how useful or efficient do you think if SITA Fast test? I think you’re talking about SITA Faster. Now, this is the go to – test glaucoma screening. Somebody comes into your OPD, everything normal, there is nothing else that is abnormal and you want to scream the patient, go for 24-2c. Even in a patient who has ocular hypertension primary angle closure suspects, primary angle closure in which no field effect this damage, OCT related thinning has been noted earlier 24-2c is the test to go forward. And that is where the evidence is strongest right now.
More and more support is coming in terms of this being used to in patients with glaucoma. But yes, we need more evidence. Thank you. [Indiscernible] [01:10:53] nice to hear from you.
False positive and negatives the minimal acceptable values, now when you’re doing an epidemiological study and you have a large database. You want a pure sample. You may keep it down to 15% or 20% dollars so that will give you the purest sample and it will give you a very good statistical desire. However in practical life, before even you look at the false positive and negative, you have to look at the disk. If the inferior rhythm is the one that is damaged and you see a superior defect, forget about false positive and negative, they only mean that you need to repeat the test again. So, statistically speaking, up to even 30% is acceptable for clinical practice. When you are looking at epidemiological studies then you have to bring down the cut off, so that the sample is pure, more pure. App for correlation between visual fields animosity, it’s not an app, it’s an inbuilt structure and function convention report that is available with the latest software version. Forum that size helps and I believe it’s available for two eye as well as HFA [indiscernible] [01:12:15].
Okay, somebody who shows poor reliability after three times forget HFA that many old patients, patients with advanced glaucomous who will not do visual fields correctly. But what I will tell you is that don’t give up fast. You reassure them, you stand with them and with required. You give them the support. You tell them that you had a site likely to miss 50% of the test. You tell them that you can pause the test, take a break and we will be with you and then give up. So, that is the situation. But if you can’t do it, you’ll have to depend OCT for earlier in moderate glaucomous and these photos for the advanced glaucomous. The OCT is not going to help you with advanced glaucoma and it’s quite poor. And I said, you will have to also listen to the patient’s symptoms. Does the patient feel that deviation is decreased without a decrease in the cataract, increase in the cataract, without any retinal change? So, that is what it is.
I will skip questions on dash two things, it’s very well described in the text books. Pupil size, yes it causes a generalized depression. But as you know that glaucoma is all about localized effects so they are not usually affected. I will skip questions on octopus. I’m not sure that that that the function structure function correlation is available for octopus or not, I would like to learn from you on that aspect.
Okay, just forum program is for any machine that you have. So, if you have a slick clam camera or take this photo, OCT slick clam photographs, everything will come together on the forum. So, it’s great to see the interface when you can see this photo on the top the visual field there and the OCT there. So, we are drawn to the computer more and more to look at the images. They’re beautiful and what a pleasure it is to look at all of these together.
NK, good to, good to hear from you. Hope you’re all fine. The VFI is not calculated for 10-2, suspected glaucoma patients. Now if we are doing in stable glaucoma patients, we are doing Visual Field one senior year so a suspect requires field less than that, not more than that.
Right, I think before the Internet clashes and — okay, I think there is one interesting question here, long one. Okay. So, what I’ll do is I think there are a lot off a lot more patients. They put my email ID on the chat box and then what you can do is write to me either of these. And then I try to answer your questions or on a public forum or a forum where you can reach out to me.
Okay, could you arrange a talking into traditional visual field? I would love to, but not now. And I must, before I go, thank Orbis and thanks Cybersight. You’ve been our partners for long and it’s a great, great forum especially in these times for learning. I regularly attend these sessions and it’s a pleasure to have everyone here. And yes, I think you must always have — already have talks on interpretation of visual field. But till then, do go through the article by Ravi Thomas in Indian Journal of Ophthalmology.
Lawrence: Great. Thank you, Doctor.
Siddharth Dikshit: Right. It’s been a pleasure, Lawrence, for always being there. Thank you. And thank you everyone for attending. I hope, wish you all the best in these difficult time, stay safe, stay healthy, take care and keep learning.
July 6, 2020
3 thoughts on “Lecture: Humphrey Field Analyzer: What the Machine Does to the Patient”
Thank you sir for this wonderful online class
Due to some error in my internet connection I was unable to attend the whole webinar .. I am listening recording right now .. would you please provide me certificate ??
Hello Maria Satter,
Thank you for your email. Unfortunately, certificates are only awarded to those who attend 80% or more of the live session.
Please email us at firstname.lastname@example.org with any other questions.