ICON Webinar

Why is Retinopathy of Prematurity still a growing problem?

Susan Ostmo teaches us why Retinopathy of Prematurity (ROP) is still a growing problem in 2022. This educational session will provide clinical insight into ROP in the past, present, and future.

Webinar Transcript


00:00:01 Candice White

So without further.

00:00:02 Candice White

Ado I want to introduce Susan Osmo.

00:00:05 Candice White

She is with us today to talk about why retinopathy prematurity is still a growing problem.

00:00:11 Candice White

Susan is a research administrator and an ROP program manager at Oregon Health and Science University in Portland, OR, and she manages several multi center RFP related research projects and she's been doing this for gosh this past 12 years.

00:00:25 Candice White

Probably feels like more to sue.

00:00:28 Candice White

She's also the coordinator for the IO IROP consortium.

00:00:32 Candice White

She does research groups, and she is part of the team that has built an artificial intelligence tool that's recently been designated as a breakthrough device by the FDA for diagnosing plus disease.

00:00:46 Candice White

Sue is Susan is the acting RLP coordinator for Dorn Becker, Neonatal Intensive care unit at Oregon Health and Science University, and works on the RFP Task Force in that unit.

00:00:55 Candice White

So she helps provide clinical research data and helps provide new protocols for oxygen management and infants at risk for Rob.

00:01:04 Candice White

She's really just immersed herself in that program to help not only the infants, but the care unit as well.

00:01:10 Candice White

Susan was the only non physician member of the writing committee for the latest International Classification of RP, the 3rd edition that was recently published and I think we're going.

00:01:18 Candice White

To hear a little.

00:01:18 Candice White

Bit more about that today.

00:01:20 Candice White

So Susan?

00:01:21 Candice White

I will turn it over to you.

00:01:23 Candice White

And thank you guys so much for being.

00:01:25 Candice White

A part of this with us today.

00:01:31 Susan Ostmo

I'm Susan Ostmo

00:01:32 Susan Ostmo

Thank you, Candace.

00:01:34 Susan Ostmo

As Candace mentioned, I'm a research administrator and RP program manager at OHSU in Portland, OR.

00:01:44 Susan Ostmo

I have been involved with RP Research and imaging for about 12 years, as she mentioned.

00:01:50 Susan Ostmo

And I've seen many changes in how the disease is diagnosed.

00:01:56 Susan Ostmo

And managed during that time span.

00:02:00 Susan Ostmo

With all of the.

00:02:01 Susan Ostmo

Knowledge that the ophthalmology community has gained in the decades since discovering the disease ROP remains the number one cause of childhood blindness worldwide.

00:02:13 Susan Ostmo

So understanding the disease itself.

00:02:16 Susan Ostmo

Is just the beginning into figuring out how to help reduce the number of children that go blind every year.

00:02:25 Candice White

So with that introduction too, I think it's really important to gauge a.

00:02:28 Candice White

Starting point for.

00:02:29 Candice White

All of us who have joined today.

00:02:31 Candice White

So how would you rate your current knowledge of ROP before we start into this session?

00:02:37 Candice White

Let's just go ahead and rate that knowledge.

00:02:40 Candice White

And then you can hit submit.

00:02:47 Candice White

And we'll watch a few of those come through for just.

00:02:49 Candice White

A second and then.

00:02:50 Candice White

We'll jump back in.

00:03:00 Candice White

All right.

00:03:01 Candice White

So we'll close this poll.

00:03:03 Candice White

You can hit done on your screen if you see that poll and I'm going to go ahead and close that out and then Susan will turn you loose.

00:03:11 Susan Ostmo

OK, thanks.

00:03:12 Susan Ostmo

I'll just begin with a brief overview of the disease and how it can lead to blindness in premature infants.

00:03:20 Susan Ostmo

I know a lot of, you know, know some about ROP or a lot about ROP.

00:03:25 Susan Ostmo

But I'll just start with the beginning. In 1942, ROP was defined as a progressive disorder that was seen exclusively in premature infants of low birth weight who were provided additional oxygen at birth.

00:03:42 Susan Ostmo

Oxygen was given liberally to many babies through oxygen cots, similar to the one seen here.

00:03:48 Susan Ostmo

In this slide.

00:03:49 Susan Ostmo

And although there was an increase in the survivability of premature infants because of the introduction of these oxygen costs, the incidence of blindness in these children was increased.

00:04:02 Susan Ostmo

So when we talk to families at our NICU or at other Nic cues, we like to relate and tell them that Stevie Wonder went blind from ROP back in 1950, when doctors knew much less about the pathology or cause of the disease, let alone about any methods for treatment.

00:04:24 Susan Ostmo

The retina is one of the last things to develop in utero and doesn't usually fully mature until about 40 weeks gestational age.

00:04:34 Susan Ostmo

When there is premature exposure to oxygen, this creates abnormal growth of the blood vessels within the underdeveloped retina.

00:04:43 Susan Ostmo

So while vascularization of the retina happens over weeks to months after the babies born, regular ophthalmoscopic monitoring is required to make sure that.

00:04:54 Susan Ostmo

Severe disease doesn't occur.

00:04:57 Susan Ostmo

So how does this abnormal blood vessel growth occur?

00:05:05 Susan Ostmo

This cartoon shows normal vessel growth within the retina while in utero, but before term.

00:05:12 Susan Ostmo

Notice the depiction of smooth vessels with normal branching that are just immature and have not yet fully vascularized the entire retina, researchers began using.

00:05:24 Susan Ostmo

Mouse and rat models to help understand.

00:05:28 Susan Ostmo

And the pathology of the disease as well as how premature exposure to oxygen affects the development of these vessels and can lead to loss of vision.

00:05:39 Susan Ostmo

So this is what we've learned from those studies.

00:05:43 Susan Ostmo

Supplemental oxygen and oxygen fluctuations can delay vascular development, so here we go from a normally developing retina with immature vessels in utero to a baby being born early, and the retina exposed to oxygen prematurely, as depicted in this right.

00:06:04 Susan Ostmo

Moon and was found in the rat and mouse models of RP progression.

00:06:09 Susan Ostmo

Oxygen seems to create retinal vessel loss and stunts vessel growth.

00:06:16 Susan Ostmo

In other words, oxygen has an inverse relationship with blood vessel growth, so this is often referred to as phase one of the disease.

00:06:28 Susan Ostmo

And as the baby matures, the resulting non vascularized retina becomes increasingly metabolically active, or hypoxic.

00:06:38 Susan Ostmo

So there's lower oxygen levels as the babies getting bigger, stronger in the mouse model of ROP.

00:06:47 Susan Ostmo

Found that there was a relationship between vascular endothelial growth factor or veg F these VEGF pros.

00:06:55 Susan Ostmo

Teens and that abnormal blood vessel growth is often referred to as the phase two of the disease.

00:07:03 Susan Ostmo

So where there used to be elevated oxygen levels, there's now a reduction in oxygen levels which stimulates the increase.

00:07:13 Susan Ostmo

Of the veg.

00:07:13 Susan Ostmo

F in the avascular area of the retina.

00:07:16 Susan Ostmo

And then that produces the abnormal blood vessel growth at the point where those vessels were shunted.

00:07:24 Susan Ostmo

So then this creates further neovascularisation at that at that demarcation line.

00:07:31 Susan Ostmo

So it that that cluster of new blood vessel growth starts growing into the vitreous as seen here in the right cartoon.

00:07:40 Susan Ostmo

So if elevated levels of VEGF persist, then ROP worsens and could develop into severe disease and cause retinal detachment. If it's permitted to progress.

00:07:56 Susan Ostmo

Very quickly I'll just review how ROP is classified.

00:08:01 Susan Ostmo

When diagnosing ROP, physicians refer to a zone, a stage, and a condition of the vessels or presence of plus disease.

00:08:15 Susan Ostmo

We'll start with zone.

00:08:17 Susan Ostmo

The zone defines the location of how far peripherally the vessels have grown.

00:08:23 Susan Ostmo

There are now 4 zones recognized within the latest classification zone.

00:08:29 Susan Ostmo

One is the area of the retina closest to the optic disc in the center.

00:08:36 Susan Ostmo

And the newly defined posterior zone 2 is recognized as that area.

00:08:41 Susan Ostmo

That's about two disk diameters outside of zone.

00:08:45 Susan Ostmo

One, but it's important to note because it should still be concerning to an examiner with vessels so posterior to the optic disc.

00:08:56 Susan Ostmo

Then there as the vessels grow out, there's zone 2, and then the most peripheral zone 3, which is the outermost Crescent and closest to full vascularization.

00:09:12 Susan Ostmo

There are five stages of ROP, with stage one being the most mild and stage five being the most severe and involves total retinal detachment and blindness.

00:09:26 Susan Ostmo

This image shows stage 2 with an elevated Ridge where the white arrow is pointing and a small tuft of stage 3 where the red arrow is pointing.

00:09:37 Susan Ostmo

So if there is more than one RP stage present, then the eye is classified by the most severe stage that is seen.

00:09:49 Susan Ostmo

Now let's discuss plus disease, which is the primary indicator of treatment requiring disease.

00:09:57 Susan Ostmo

So plus disease is defined by appearance of dilation and tortuosity of the retinal vessels and pre plus.

00:10:08 Susan Ostmo

Is slightly less, but still has abnormal vascular dilation and tortuosity.

00:10:15 Susan Ostmo

The posterior vessels within zone one are the vessels that are examined when you are considering the presence of plus and pre plus.

00:10:26 Susan Ostmo

The condition of the vessels is used as this marker for ROP that requires treatment, but often even world experts disagree about what constitutes plus disease and will continue to talk about that as we go along.

00:10:45 Susan Ostmo

So now let's get to the topic of the day.

00:10:48 Susan Ostmo

Why is ROP a growing problem when we understand so much more about the pathology of the disease?

00:10:56 Susan Ostmo

And in this day and age of advanced technology?

00:11:00 Susan Ostmo

As mentioned earlier, medical advances have drastically increased infant survivability when they're born prematurely.

00:11:08 Susan Ostmo

Babies born up to 17 weeks early can survive because of supplemental oxygen, and so more babies are needing exams.

00:11:20 Susan Ostmo

ROP is a growing international problem, especially in areas of the world that do not have specialized Nic cues or have limited access to resources.

00:11:32 Susan Ostmo

With more babies at risk of severe ROP and that are in need of routine screening, there are also not enough specialists to diagnose the disease adequately.

00:11:43 Susan Ostmo

There is high medical legal liability compared with the amount that is compensated to physicians for exam.

00:11:52 Susan Ostmo

And on top of all of this, the disease itself is difficult to diagnose.

00:11:57 Susan Ostmo

As mentioned earlier, even so-called world experts can disagree with one another on what is deemed treatment requiring disease.

00:12:07 Susan Ostmo

So let's breakdown all of these issues.

00:12:13 Susan Ostmo

First of all, the three there's been 3 epidemics of blindness due to RP that have been described. The 1st and 2nd epidemics both occurred in high income countries initially in the 1940s and 50s because of preterm infants being exposed to unmonitored 100%.

00:12:33 Susan Ostmo

Supplemental oxygen in order to help them survive.

00:12:37 Susan Ostmo

The last 30 years have been experiencing the third epidemic, which has features of the 1st and 2nd epidemics and is occurring in countries and regions that have scaled up their neonatal intensive care, but even higher income countries are still seeing a continued increase in infant surviving.

00:12:57 Susan Ostmo

The earlier and earlier ages creating larger numbers of babies in need of RP screening.

00:13:05 Susan Ostmo

And in 2010, it was estimated that 32,300 pre term infants worldwide became blind or visually impaired from ROP with the highest number in East, Southeast Asia and Pacific region.

00:13:22 Susan Ostmo

You can see from these.

00:13:24 Susan Ostmo

Small pie charts depicting various regions and percentages of survival, but with blindness or severe visual impairment.

00:13:33 Susan Ostmo

It's much higher in the lower to middle income countries that medium purple slice reflects those numbers of children.

00:13:42 Susan Ostmo

Who go blind within that region?

00:13:45 Susan Ostmo

Higher income countries still have a small percentage of children going blind, but lower and middle income countries have the largest percentages.

00:13:55 Susan Ostmo

The low to middle income countries, there's a disparity in relevant ROP screening guidelines or policies resulting in the use of this unmonitored oxygen.

00:14:08 Susan Ostmo

In these settings where neonatal care is of suboptimal quality, it leads to these infants who are larger and more mature, developing severe and aggressive ROP.

00:14:23 Susan Ostmo

The next issue in the growing ROP program is too few physicians to perform screening exams with more and more babies at risk of ROP, there is a lack of doctors wanting to specialize in RP management.

00:14:39 Susan Ostmo

So as the numbers of infants are increasing, the number of physicians to examine them.

00:14:44 Susan Ostmo

Is decreasing.

00:14:46 Susan Ostmo

There are a few, a few reasons that the decrease in positions is happening, one being that from an ophthalmology standpoint.

00:14:53 Susan Ostmo

It ROP care is complex. It spans multiple care settings and providers, including those in Nicu's nurseries and outpatient clinic settings. This requires coordination and communication between providers and ancillary.

00:15:13 Susan Ostmo

Staff and most importantly, effective communication with patients, family members to impress upon them the importance of continued follow up.

00:15:24 Susan Ostmo

So often providers have to travel too far away.

00:15:28 Susan Ostmo

Nic used to perform exams, coordinating this extremely important follow up is also time consuming.

00:15:37 Susan Ostmo

This is worse in low to middle income countries because it's increasingly growing problem.

00:15:44 Susan Ostmo

With so many kids surviving at earlier ages, aside from the time consuming coordination, the RP exam itself is very difficult to perform.

00:15:56 Susan Ostmo

It takes a very patient and skilled person to perform an RP exam with babies trying to stop breathing, and we know that babies and nurses and families get stressed during our OP exams.

00:16:10 Susan Ostmo

But physicians are under a lot of stress also trying to perform a thorough exam on an already fragile.

00:16:17 Susan Ostmo

Infant is far from easy.

00:16:19 Susan Ostmo

And for many trainees trying to decide what area to specialize in, there are other areas of ophthalmology that are much less stressful and easier to take on and specialize in other than ROP.

00:16:38 Susan Ostmo

There is also a lack of ROP training for residents and fellows in ophthalmology.

00:16:44 Susan Ostmo

Many ophthalmology trainees report that ROP exams done during their training were performed without an attending directly supervising at the time of the examination.

00:16:56 Susan Ostmo

Questions, in other words, a fellow examines a baby alone, but the attending follows with the examination at.

00:17:02 Susan Ostmo

A different time and.

00:17:04 Susan Ostmo

These separate examinations for ROP result in increased stress for already sick infants as well as logistical challenges with respect to coordination of care in the NICU.

00:17:17 Susan Ostmo

Trainees examining under direct supervision of attendings minimize these issues and report that the direct supervision provides additional learning opportunities with regard to clinical findings, treatment recommendations, and even advice given regarding examination techniques.

00:17:38 Susan Ostmo

So these supervised trainings are the most ideal.

00:17:41 Susan Ostmo

But the supervising physicians report a lack of willingness to have every baby endure 2 exams, so they prefer very few of these babies to do the double exam.

00:17:56 Susan Ostmo

Hands on training is always the most beneficial.

00:18:01 Susan Ostmo

But there's also a lack of external sources for training to diagnose ROP.

00:18:06 Susan Ostmo

This slide actually shows a student training module created by the Iraq consortium back in 2013, and it's been tested here in the States and utilized internationally as a tool for additional training. The Iraq consortium continues to improve these external sources for training, but.

00:18:26 Susan Ostmo

Nothing compares to hands on training, which is often difficult to obtain.

00:18:37 Susan Ostmo

Another huge hurdle to the decrease in position specializing in ROP is the medical legal liability involved.

00:18:44 Susan Ostmo

ROP is one of the most costly medical legal areas within ophthalmology, which has made it more challenging to find qualified physicians willing to take on this area of practice.

00:18:57 Susan Ostmo

That and reimbursement for providing RP screening tends to be less than that which ophthalmologists could generate, providing outpatient clinical care or performing ophthalmic surgery.

00:19:09 Susan Ostmo

As mentioned earlier, the process tends to be time consuming, often requiring physicians to travel to various.

00:19:16 Susan Ostmo

Hospitals away from their regular work site and there's additional cost to the practitioner incurred for compensating administrative staff to coordinate this outpatient care.

00:19:31 Susan Ostmo

Poor reimbursement and the risk of multi, multi $1,000,000 malpractice settlements have made providing this service pretty undesirable.

00:19:39 Susan Ostmo

Additionally, documentation of Progressive ROP has historically not been great.

00:19:47 Susan Ostmo

This slide shows how ROP has been documented in the past, which is not great for RECO.

00:19:53 Susan Ostmo

Starting making record of an exam and trying to prove medical legal liability.

00:20:00 Susan Ostmo

So use of imaging systems have helped enormously in being able to track and record disease for progression or regression as well As for liability purposes and hopefully imaging for OP will become the standard of care.

00:20:15 Susan Ostmo

Everywhere and will help support physicians making record of what they see as they perform these exams.

00:20:22 Susan Ostmo

Week to week.

00:20:27 Susan Ostmo

Another huge problem with ROP is that even world experts disagree on what warrants treatment requiring disease.

00:20:37 Susan Ostmo

So if experts disagree, how are undertrained ophthalmologists supposed to manage disease?

00:20:44 Susan Ostmo

Well, for their patients.

00:20:46 Susan Ostmo

This slide shows a study that was performed comparing various images with varying degrees of vascular severity that were evaluated by ROP experts for plus disease diagnosis.

00:21:01 Susan Ostmo

The main finding from the study is that inter expert agreement of plus disease is imperfect.

00:21:09 Susan Ostmo

Representative images were shown to 22 expert participants, and I'll just kind of go through and explain these and the first image was classified as.

00:21:22 Susan Ostmo

Normal by all 22 experts, the second image was classified as plus by two experts and pre plus by 9 experts and normal by 11 experts. So quite a disparity. The upper right image is classified as plus by 1.

00:21:42 Susan Ostmo

Expert pre plus by 16 experts and normal by 5 experts and then so on so you can see that there's a disparity other than the lower center one which was 100% plus by all experts. There is definitely a discrepancy in how.

00:22:00 Susan Ostmo

These experts are classifying plus disease.

00:22:05 Susan Ostmo

There's several potential explanations for the variability in the diagnosis.

00:22:11 Susan Ostmo

Some may pay more attention to tortuosity versus dilation, or some experts are more aggressive in their treatment plans where others are more conservative.

00:22:23 Susan Ostmo

But this inconsistency amongst those who are the most experienced makes it difficult to come to a consensus on when is best to treat.

00:22:38 Susan Ostmo

In the interest of always trying to improve ROP diagnosis and management and to help with more expert agreement, the International Classification, Classification of retinopathy of Prematurity Committee, also known as Eye Crop Committee, was formed several decades ago and is a group of world.

00:22:58 Susan Ostmo

Experts that changes, you know, as people retire and come into the.

00:23:03 Susan Ostmo

The field, but there are world experts on ROP who write a consensus statement happens about every 15 years.

00:23:12 Susan Ostmo

That creates a standard nomenclature for classification of the of the disease and some of that we've already discussed.

00:23:19 Susan Ostmo

When we were describing zone and stage and plus.

00:23:24 Susan Ostmo

The I Crop Committee came up with these terms and definitions when they initially published in 1984, and then it was expanded. The classification was expanded again in 1987 and revisited in 2005.

00:23:40 Susan Ostmo

The classification was updated most recently in 2021 in an article presenting 1/3 revision. Or we like to call it the I crop the.

00:23:52 Susan Ostmo

The group felt revisions were required for many reasons, some of which involved concerns about subjectivity in critical elements of disease classification, including expert disagreement with plus disease, as well as recognition that patterns of ROP in some regions of the world.

00:24:13 Susan Ostmo

Do not fit neatly into that current classification system.

00:24:22 Susan Ostmo

In their deliberations to update disease classification, the eye crop 3 Committee conducted some image reading exercises in 2019. The 34 world experts in ROP were shown the same 30 images and were asked to classify each image as normal vessels pre plus vessels.

00:24:42 Susan Ostmo

Or plus vessels.

00:24:44 Susan Ostmo

They found that they were reasonably consistent when images were extremely severe and very mild, but the inconsistent inconsistencies came in the areas bordering between normal and pre plus and also between pre plus and plus.

00:25:03 Susan Ostmo

You can see that this slide.

00:25:06 Susan Ostmo

In this slide that the most extreme plus images are red and we're consistently ranked by the 34 experts as severe at the top and that the most normal, which were represented as green, we're consistently ranked by the experts as mild. There is wide variability in.

00:25:28 Susan Ostmo

The classification by ROP experts when it comes to these between areas, but those are the areas that are very important in terms of determining closer follow up or even whether it's a borderline treatment situation.

00:25:46 Susan Ostmo

These variations, as we talked about, could occur because experts have different personal cut points for the amount of vascular abnormality required for plus disease, and the committee recommended that terms like pre plus and plus.

00:26:06 Susan Ostmo

Should continue to be used, but emphasize that these terms represent a continuous spectrum of retinal vascular changes.

00:26:21 Susan Ostmo

The IROP consortium of researchers initially brought about this idea in an article in ophthalmology by Campbell ET al. In 2016.

00:26:33 Susan Ostmo

As mentioned, the wide variability and classification of plus disease by the ROP experts occurs because experts have a different.

00:26:42 Susan Ostmo

Poise for the amounts of vascular abnormality.

00:26:46 Susan Ostmo

And this is has important implications for research, for teaching and for patient care, and suggests that continuous plus disease severity score may reflect more accurately the behavior of expert or OP clinicians.

00:27:06 Susan Ostmo

And that it may better standardize classification in the future.

00:27:13 Susan Ostmo

Because of these findings, and because of their own variability in the classification exercise is one of the major updates in the latest I crop 3 is a recognition.

00:27:24 Susan Ostmo

That plus is a continuous spectrum from normal to pre plus to plus.

00:27:29 Susan Ostmo

With these sample images demonstrating this.

00:27:32 Susan Ostmo

Range and this figure demonstrates gradings of the spectrum by the members of the committee, and although Gradings along the spectrum may vary amongst the observers, that there was better agreement on either end of the spectrum, but that all of the experts.

00:27:53 Susan Ostmo

Were able to place the same images in order of severity, whether they called it normal or pre.

00:28:00 Susan Ostmo

Plus there is definite agreement and when the disease is progressing.

00:28:07 Susan Ostmo

This has important importance in clinical practice for assessing, you know, whether disease is getting better or worse.

00:28:18 Susan Ostmo

So what does all this mean?

00:28:20 Susan Ostmo

Without consistent diagnosis of plus disease, which is the most important marker for physicians in determining the need for treatment?

00:28:29 Susan Ostmo

There will continue.

00:28:29 Susan Ostmo

To be some babies that are not treated in time and others who are treated potentially unnecessarily.

00:28:42 Susan Ostmo

Given all of this, where are we going in the future in order to improve our OP diagnosis and timing of treatment?

00:28:53 Susan Ostmo

Firstly, the use of artificial intelligence in helping to diagnose ROP has been an area of research for a long time, but now there is a tool that has been fast tracked with the FDA as a breakthrough device and validation of that system is currently underway.

00:29:12 Susan Ostmo

Talk more about that in a minute. Secondly, because of the lack of ophthalmologists trained in diagnosing ROP's, telemedicine programs are growing worldwide and.

00:29:24 Susan Ostmo

Will hopefully help.

00:29:26 Susan Ostmo

To fill those gaps in care, thirdly, risk models are being studied.

00:29:32 Susan Ostmo

With the hope of eventually being able to reduce the number of OP exams necessary on lower risk babies.

00:29:46 Susan Ostmo

OK, the first of these, where are we going?

00:29:48 Susan Ostmo

Topics is the advance of artificial intelligence in ROP.

00:29:55 Susan Ostmo

Because plus, disease is the major determinant in diagnosing treatment requiring ROP efforts have been made to quantify the vascular changes in ROP using the features of dilation and tortuosity.

00:30:11 Susan Ostmo

Initially, computer systems were trained on retinal images with manually traced vessels, and then these systems could automatically identify and trace blood vessels on their own and quantify the amount of dilation and tortuosity within the retinal image.

00:30:30 Susan Ostmo

So in 2018, Brown ET al. Reported the results of a fully automated deep learning based system for automated diagnosis of plus disease.

00:30:42 Susan Ostmo

The system essentially was trained on more than 5000 images. All those images were assessed by a consensus diagnosis of three independent image graders, along with an ophthalmoscopic diagnosis to create one reference standard.

00:31:03 Susan Ostmo

The reference standard was provided for all 5000 plus images and the system was trained on those.

00:31:11 Susan Ostmo

The machine learning systems they learn the features that best correlate with the input image and with the reference standard diagnosis of the experts.

00:31:21 Susan Ostmo

So it takes what the vessels look like.

00:31:25 Susan Ostmo

It takes the expert diagnosis and it learns what it's supposed to say in terms of normal pre plus or plus and give it a quantifiable diagnosis.

00:31:38 Susan Ostmo

This fully automated algorithm, diagnosed plus disease in RP.

00:31:42 Susan Ostmo

With comparable or better.

00:31:44 Susan Ostmo

Accuracy than human experts when it was tested.

00:31:52 Susan Ostmo

This AI algorithm was further developed to produce that quantifiable measure of ROP severity.

00:31:59 Susan Ostmo

The IROC consortium developed an automated ROP vascular severity score, through these computer science methods and came up with a score that.

00:32:08 Susan Ostmo

Ranged from 1:00.

00:32:10 Susan Ostmo

To not.

00:32:11 Susan Ostmo

This quantitative score is consistent and reflects disease progression, so each of these skeleton images represent a number of 1 to 9.

00:32:24 Susan Ostmo

So similar to getting your blood pressure numbers, the vascular severity.

00:32:31 Susan Ostmo

Score would give a number from 1:00 to 9:00 and is a tool that may be helpful for physicians as assistive diagnosis in determining when it's appropriate timing for treatment or retreatment.

00:32:49 Susan Ostmo

One common criticism for AI algorithms such as this are that there's.

00:32:56 Susan Ostmo

The acceptance by physicians and patients may be low because of the inability to explain how the algorithm arrived at a conclusion.

00:33:09 Susan Ostmo

These AI systems are often thought of as black boxes, where there is a lot of unexplained learning that happens and that makes.

00:33:18 Susan Ostmo

Old school medical people, understandably nervous.

00:33:22 Susan Ostmo

But these systems are becoming more and more accurate.

00:33:25 Susan Ostmo

And despite the Inexplicability can help better quantify and diagnose our OP.

00:33:33 Susan Ostmo

And as AI enters clinical medicine, there is increasing awareness of the need to adjudicate liability from care decisions informed by the AI.

00:33:45 Susan Ostmo

Victims and to this end, there is a distinction between autonomous and assistive AI systems.

00:33:53 Susan Ostmo

So in autonomous systems, decisions are based solely on the output of the AI system, and in assistive systems the output is used to aid a clinical diagnosis.

00:34:06 Susan Ostmo

By a physician.

00:34:07 Susan Ostmo

And the FDA is rapidly innovating their methods for evaluating to ensure safe implementation of these technologies and clinical care.

00:34:18 Susan Ostmo

So as the technologies become more commonplace, the regulatory requirements will likely continue to evolve, as will be medical.

00:34:27 Susan Ostmo

Little implications.

00:34:34 Susan Ostmo

The second of the where are we going with all of these issues topics is telemedicine the ability to easily image a neonatal retina paved the way for telemedicine to provide a more efficient method to screen at risk.

00:34:53 Susan Ostmo

These there are now multiple examples of successful telemedicine.

00:35:00 Susan Ostmo

Programs in the United States and around the world.

00:35:04 Susan Ostmo

Especially in regions where there are too few trained or willing ophthalmologists to manage ROP screening and treatment, telemedicine can allow a single provider to screen babies over a larger geographic area.

00:35:22 Susan Ostmo

And for example here in the state of Oregon, there are many remote areas where medical care as specialized as ROP screening is not available, and there are only a handful of provider.

00:35:34 Susan Ostmo

In the entire state.

00:35:36 Susan Ostmo

Babies need to be transferred largely to one or two high acuity facilities to get appropriate ophthalmic care.

00:35:44 Susan Ostmo

With addition of more of these telemedicine programs, more remote facilities will be able to provide more comprehensive care for premature babies, and families will be able to stay closer to home and not be inconvenienced with a six hour drive just to have their baby screened for ROP.

00:36:04 Susan Ostmo

It is the.

00:36:04 Susan Ostmo

Hope that after validation of the IROC deep learning system with the FDA that the technology can be available and all fundus cameras could be equipped with the software to produce this vascular severity score for retinal images and this could help telemedicine.

00:36:24 Susan Ostmo

Programs in their ability to better manage our OP care for infants.

00:36:35 Susan Ostmo

The third topic.

00:36:36 Susan Ostmo

Of where we're going is this ROP risk prediction.

00:36:42 Susan Ostmo

We know that RP screenings are essential service in Nic use.

00:36:46 Susan Ostmo

However, the current risk models subject infants to multiple physiologically stressful exams.

00:36:54 Susan Ostmo

Umm, so last year.

00:36:57 Susan Ostmo

The IROC consortium researchers studied 2 separate data sets of patients who were screened multiple times for ROP and fundus images were taken at all of the exams.

00:37:09 Susan Ostmo

The images were analyzed by the previously discussed the IROP deep learning algorithm.

00:37:19 Susan Ostmo

Cut points at within a window of 32 to 34 PM, a post post, menstrual age imaging window.

00:37:27 Susan Ostmo

A model for prediction of treatment requiring ROP was trained and optimized, and the researchers were able to demonstrate that analysis of the images at that single exam 32 to 34 weeks detected 100% of all infants who eventually.

00:37:47 Susan Ostmo

Developed treatment requiring RP and it also is able to detect to detect more than half of those who did not develop severe disease at all.

00:38:00 Susan Ostmo

This table highlights those kids that were eventually treated and in both data sets all eventually treated kids were predicted to need treatment through analysis of those images taken at 32 to 34 weeks.

00:38:14 Susan Ostmo

Although the system did identify many kids, that didn't end up needing treatment, it did catch about half of those who.

00:38:23 Susan Ostmo

Wouldn't develop disease.

00:38:24 Susan Ostmo

So what does this mean?

00:38:27 Susan Ostmo

Obviously for the for the research needs to be done in this area, but the implementation of a model such as this could lead to significantly fewer ROP exams.

00:38:38 Susan Ostmo

For lower risk infants.

00:38:41 Susan Ostmo

It could be.

00:38:42 Susan Ostmo

A better use of ROP screening resources and provide an earlier recognition of treatment requiring ROP.

00:38:52 Susan Ostmo

These are important because you know, we all would love less eye exams.

00:39:04 Susan Ostmo

With this risk model, there are limitations of the the prediction model.

00:39:10 Susan Ostmo

That one is that it was trained on a data set of infants in the United States, and while it works well with that particular demographic and would likely translate well to other countries with similar demographics.

00:39:25 Susan Ostmo

The tool would have to be separately trained on data sets from low to middle income countries where ROP often occurs in older, heavier babies.

00:39:36 Susan Ostmo

Future work to validate this concept in low to middle income countries where potential added value may be even greater.

00:39:45 Susan Ostmo

Given the increasing prevalence of the disease and scarcity of resources.

00:39:50 Susan Ostmo

With the goal of reducing and eliminating blindness due to ROP.

00:39:54 Susan Ostmo

So currently we have data sets from Mongolia and Nepal and India that we are looking at and trying to validate or risk prediction model for those demographics and then?

00:40:10 Susan Ostmo

Will hopefully lead to eliminating a lot of the eye exams that actually need to occur in those regions.

00:40:23 Susan Ostmo

So blindness from ROP is avoidable, but currently still a global health problem.

00:40:32 Susan Ostmo

Current research to develop artificial intelligence systems is promising, and implementing this technology into existing imaging systems could be a solution to this worldwide issue.

00:40:46 Susan Ostmo

There are several potential challenges to ensuring that.

00:40:49 Susan Ostmo

Every at risk baby is.

00:40:51 Susan Ostmo

Diagnosed accurately and in a timely manner.

00:40:55 Susan Ostmo

Besides the wide disparities worldwide in the distribution of ophthalmologists between rural and urban settings and between countries, the diagnosis of RP is based on this subjective assessment of a disease.

00:41:10 Susan Ostmo

That disease severity and it will, it's well established that.

00:41:15 Susan Ostmo

There is wide variability for all three components in diagnosing the disease.

00:41:21 Susan Ostmo

It's the most examiners don't routinely perform photography at the time of examination, which hinders their ability to objectively make comparisons across serial exams or even between other examiners.

00:41:38 Susan Ostmo

So the adaptation of fundus photography as a standard of care.

00:41:42 Susan Ostmo

Can help with this objectivity and with the addition of AI software to produce a vascular severity score in assisting with the diagnosis as well as implementing some of these risk models can all help.

00:41:58 Susan Ostmo

Provide accurate screening, timeliness of treatment and hopefully begin to alleviate this growing worldwide problem of ROP.

00:42:12 Candice White

Thank you so much, Susan.

00:42:13 Candice White

That was a wonderful presentation.

00:42:16 Candice White

So the last poll that I want to launch for everyone real quickly.

00:42:21 Candice White

Is now that we have had all of this information we went through.

00:42:27 Candice White

You know why is ROP still a growing problem?

00:42:30 Candice White

Let's see if we can gauge our knowledge following the information.

00:42:44 Candice White

So we'll take a moment here for everybody to rate.

00:42:48 Candice White

How they feel their knowledge is after the presentation.

00:42:53 Candice White

Give it a few minutes for everyone to respond.

00:43:03 Candice White

Next and you guys can click done.

00:43:04 Candice White

When you're finished with that.

00:43:08 Candice White

And then the.

00:43:08 Candice White

Last thing we'll do is we'll open this session.

00:43:11 Candice White

Up for any.

00:43:13 Candice White


00:43:16 Candice White

In the chat box.

00:43:17 Candice White

So if you're on a phone or a computer system, you may find your chat there may be 3 dots at the bottom, which would be a more tab.

00:43:24 Candice White

If you click on that, it can open up a chat box.

00:43:27 Candice White

If not, we've got chat on the side of the screen.

00:43:31 Candice White

For those of us that are joining us on the computer, and I had a couple questions submitted to me, so we can post those to Susan and let her expertise answer some of our.

00:43:43 Candice White

Wonderful questions.

00:43:44 Candice White

So Susan, I know one of.

00:43:45 Candice White

The questions that we had pop up is.

00:43:48 Candice White

Do you see?

00:43:50 Candice White

The additions.

00:43:52 Candice White

Of cameras in the care units being important now.

00:43:56 Candice White

As we see.

00:43:57 Candice White

The growth of AI, kind of on our heels.

00:44:00 Candice White

So do you kind of feel like now is the time to be placing those systems so that when AI starts to take off, we're ahead of the game?

00:44:10 Susan Ostmo

I think what we're finding is, you know, for about a decade.

00:44:17 Susan Ostmo

We have utilized.

00:44:21 Susan Ostmo

Imaging systems in our NICU and we have the retina specialist or the pediatric ophthalmologist right there with us.

00:44:32 Susan Ostmo

So it's not, that's not for a telemedicine purpose.

00:44:36 Susan Ostmo

What we use it for is to track disease.

00:44:42 Susan Ostmo

So beyond the AI tracking progression or regression of the disease we feel like is most.

00:44:50 Susan Ostmo

Beneficially done by taking these images and having a recording of what is happening for each patient.

00:44:58 Susan Ostmo

Beyond that, I think that having.

00:45:02 Susan Ostmo

Units get used to there's a learning curve and in taking these images and because we have a lack of positions.

00:45:14 Susan Ostmo

And examiners building these telemedicine programs now before the AI is actually available is actually going to be super helpful, because the learning curve.

00:45:30 Susan Ostmo

In terms of getting quality images is you know it's a little bit time consuming so.

00:45:38 Susan Ostmo

Getting units used to fund this imaging now is probably going to be most beneficial for when that technology is ready.

00:45:49 Candice White

That makes sense.

00:45:50 Candice White

I think it's so awesome.

00:45:51 Candice White

To hear you say that.

00:45:53 Candice White

You've got cameras placed within your units that really aren't telemedicine driven.

00:45:58 Candice White

I think sometimes we can get it in our minds that, oh, if I place a camera in my system that I have to go full telemedicine and we have seen so many successful programs.

00:46:06 Candice White

Kind of grow.

00:46:08 Candice White

Into large telemedicine, meaning they still have their ophthalmologist right at their side, almost involved and then we slowly start to.

00:46:15 Candice White

To build a bigger.

00:46:16 Candice White

Gap to help both sides provide the best care possible.

00:46:20 Candice White

So I think that's a really aha moment to hear that they have so many options.

00:46:25 Susan Ostmo

Well, the one other advantage.

00:46:28 Susan Ostmo

I think of keeping a system in the rounding.

00:46:35 Susan Ostmo

Protocol is that when you are training residents and fellows.

00:46:42 Susan Ostmo

There's nothing like having them see on a camera what they're supposed to be seeing using the indirect. So we find that at KCI, our residents and fellows are much.

00:46:59 Susan Ostmo

That are prepared in going out and being able to diagnose the disease than a lot of other trainees.

00:47:09 Candice White

Another question that we had pop up was why was posterior Zone 2 added?

00:47:14 Candice White

Like why did we see a value to kind of break up those more interior zones.

00:47:21 Susan Ostmo

I think there is an equipoise for a lot of physicians to I think it has to do with treatment.

00:47:30 Susan Ostmo

There's a couple different types of treatment and I think that a lot of physicians really struggle with.

00:47:38 Susan Ostmo

Lasering, a baby that has treatment requiring disease with disease, still posterior, still so posterior, so you may be out of zone one, but you're so close to the area of central vision and it's a lot of real estate to have to laser.

00:47:55 Susan Ostmo

So I think that the crop committee.

00:47:58 Susan Ostmo

Really wanted to recognize the importance of that still posterior area because I think it gave physicians.

00:48:09 Susan Ostmo

A better like, hey, yeah, this is still OK to inject, even though you're in zone 2.

00:48:17 Susan Ostmo

But buy yourself some time to let those vessels grow out a little bit.

00:48:22 Susan Ostmo

And I think a lot of it had to do with, you know, there was this standard, you know.

00:48:29 Susan Ostmo

10 years ago, like you, you don't inject when you're into Zone 2 and babies were getting lasered with.

00:48:37 Susan Ostmo

You know, a lot of retina still left to develop.

00:48:41 Susan Ostmo

So I mean I think that that.

00:48:43 Susan Ostmo

Was kind of the the.

00:48:45 Susan Ostmo

Main reason for that?

00:48:46 Susan Ostmo

But also to recognize, you know, Zone 2 is a huge area.

00:48:50 Susan Ostmo

So are we are we peripheral zone 2, are we are we posterior zone 2?

00:48:57 Susan Ostmo

I mean, and this is all it, it's kind.

00:48:58 Susan Ostmo

Of like you.

00:48:59 Susan Ostmo

Know plus even zone is a spectrum.

00:49:02 Susan Ostmo


00:49:03 Susan Ostmo

I mean, there's there's distances out from the optic disc, but it's all a spectrum of disease.

00:49:11 Susan Ostmo

We even talked about stage as a spectrum.

00:49:14 Susan Ostmo

You know Pete Campbell and I talk like it's a 1.2 or it's a 2.3. I mean we it.

00:49:22 Susan Ostmo

And so it is.

00:49:23 Susan Ostmo

It it's all a.

00:49:24 Susan Ostmo

Spectrum, but I think that we wanted to really recognize that that posterior zone too is much more important than to to recognize.

00:49:35 Candice White

So we've had another question posed when we talked about the difficulty to diagnose some cases of OROP, especially if they're not clear on that plus side or normal case.

00:49:46 Candice White

So other than an ophthalmoscope and a fundus photo, what other tools does a physician have to diagnose ROP?

00:49:54 Candice White

Is it FA?

00:49:55 Candice White

Do we see?

00:49:56 Candice White

CT on the horizon, even maybe OCD angiography, this is everyone's favorite subject is to.

00:50:03 Candice White

Say when are we gonna say OCD and P?

00:50:07 Susan Ostmo

Such a timely question.

00:50:09 Susan Ostmo

I work with a.

00:50:10 Susan Ostmo

Group that is.

00:50:11 Susan Ostmo

Building Candace is seeing this camera.

00:50:15 Susan Ostmo

It shows on Foss images, but it's an Oct and Octa and it's, you know, infrared light.

00:50:24 Susan Ostmo

So it's not as.

00:50:25 Susan Ostmo

You know.

00:50:27 Susan Ostmo

Stressful to babies, it has a wide field of view and it's it really captures much more quickly.

00:50:35 Susan Ostmo

You know, a wide range.

00:50:37 Susan Ostmo

So yeah, Oct that's I think where.

00:50:42 Susan Ostmo

This is going it is.

00:50:51 Susan Ostmo

You know, obviously a lot of validation needed in in cameras like this, but.

00:50:58 Susan Ostmo

You know, it's it's important to recognize that imaging.

00:51:03 Susan Ostmo

Even these these fundus cameras.

00:51:07 Susan Ostmo

It's they're easy to find, to just be able to get in and see posterior pull images, and that's the most important piece in trying to diagnose.

00:51:20 Susan Ostmo

Plus disease is getting that posterior pull image.

00:51:25 Susan Ostmo

And yeah, Oct is on the horizon.

00:51:29 Candice White

So we'll put you.

00:51:30 Candice White

On the spot here, how far away do you think we are from the AI software?

00:51:34 Candice White

I know that we've been.

00:51:35 Candice White

Pushing it the FDA.

00:51:36 Candice White

Is is seeming very accepting.

00:51:39 Candice White

So when do we think we'll start to see this?

00:51:45 Susan Ostmo

I think.

00:51:47 Susan Ostmo

Probably within the next year.

00:51:53 Susan Ostmo

Have a data set that we are so we we have the the algorithm.

00:52:00 Susan Ostmo

We just have a separate data set that we are validating it on and I think we're going to see within the next year some application submitted and.

00:52:14 Susan Ostmo

Acceptance of.

00:52:15 Susan Ostmo

A1 to 9 scale and it'll probably be an assistive tool, not an autonomous, but I think that that will be a huge benefit to physicians in deciding.

00:52:32 Susan Ostmo

No treatment in the future.

00:52:36 Candice White

I know we have talked about this.

00:52:37 Candice White

Anybody who's spent time with me or or probably Susan in the past knows we've we've kind of talked about how OP is living a little bit.

00:52:44 Candice White

In the past, we've seen so many advancements for diseases and adults like diabetics and you know, macular generation.

00:52:52 Candice White

So I'm super excited to see ROP kind of jump into the gang and really start introducing these new diagnosing options and photographic tools to document what we're seeing.

00:53:05 Candice White

So we are at the top of.

00:53:08 Candice White

The hour I'm so.

00:53:10 Candice White

Excited to be able to present information like this.

00:53:12 Candice White

I'm so thankful, Susan, that you took the time to spend with us today and teach us more about why our OOP is still a growing program.

00:53:19 Candice White

For those of.

00:53:20 Candice White

You that have joined, we will have these sessions recorded and available to you.

00:53:24 Candice White

You'll be getting some emails and some follow up information.

00:53:26 Candice White

As well as an evaluation so that we can continue to provide the best sessions possible, I can't thank you guys enough for spending the morning with us and we look forward to seeing you on a future East seminar.