Monika ([00:00].128)

I think any pathologist who has gone through the rigorous training procedures which we go through for so many years of our life, they know the importance of color in pathology. So it is not just about pink and blue in age and eem, but also the beautiful different stains and colors. So I think that these colors really produce valuable diagnostic information and helps the pathologist to identify disease and to determine the effects.

Tom: Things are moving so fast, so much more insights about the importance of color, regulatory changes that are happening. So it would not be wise if you invest today not to make sure that your entire imaging chain is properly color managed. This is technically possible, it will not be perfect. Every step you take you will move towards more color consistency.

Aleks: Welcome my digital pathology trailblazers. I’m super excited about the topic of today’s episode because it’s about color calibration in digital pathology. This is one of those topics that everybody agrees it is important, but few know how to actually do it. Few know like how to start with it and ask any pathologists making the digital transition and they’ll tell you that they’ve heard color matters, right? This is what we work with. Faithful representation of color is critical.

But when it comes to practical guidance, there’s a little confusion and we’re asking pathologists to move from microscope where they control everything. They control the light source. They see the tissue in a physical form to computer screens where colors can shift, degrade, and vary between displays. So even though we’re doing that, the choice of monitor for the lab is based on the price rather…

than color fidelity, often without understanding the potential consequences. And this is what we’re going to be talking about today. I have two amazing experts, Dr. Monica Lamba Saini, Global Translational Pathology Leader at ADC Therapeutics, who brings her clinical user perspective from different organizations, including CROs and biopharma, and Tom Kimpe, VP of Technology and Innovation at Barco Healthcare.

Aleks ([02:11].16)

who has the engineering expertise to explain what’s really happening with color in digital pathology. So glad to have you on the podcast. Welcome, Monica. Welcome, Tom. How are you today?

Thank you so much for the invite, Alex.

Of course, and Monica, it’s not our first rodeo here. You were already guest on my podcast. But for those who have not heard the previous episodes, and we always start with the guests. So Monica, I’m going to ask you to give a little bit of your background and your perspective that you’re going to be covering the user perspective in this color discussion today. And then Tom, I’m going to ask you to do the same after Monica is done.

Thank you so much again for the invite, both Alex and Tom. At the outset, I would like to say that I have no commercial affiliation to Barco. The perspective which I’m going to share today is across different organizations working on different screens, including Barco, and the kind of experience you have when you have calibrated monitors and when you do not work on calibrated monitors.

Alex, you remember the last time when we spoke, I had discussed about my journey into the fascinating world of clinical trials. So trained as an MD pathologist and then with a PhD, which incorporated both oncopathology, molecular pathology and digital pathology, I believed clinical trials were the perfect culmination for both of my skills, diagnostic and research.

Monika ([03:42].572)

I spent almost nine years working with CROs in different countries and then decided to make the switch to the other side, which is the biopharma, when ADCT contacted me for the present role. To me, pathology is not just about examining human tissue. It’s about understanding the intricate workings of the human body at both cellular and sometimes molecular levels.

So I believe as a pathologist working on drug development and clinical trials, I’m able to do that. And my work focuses on translating, on advancing translational research strategies, leveraging digital pathology and integrating biomarkers to drive novel therapeutics from discovery to clinical development.

I think clinical trial work is a keyword because, and I’m going to link to our previous podcast because this is where digital pathology actually telepathology originated. And here this, this topic of color faithfulness across different places where you look at them is crucial as well. Tom, if you could give a few words about your background and also like in general.

from an engineering standpoint, why is color so much more critical in pathology than other medical fields like radiology or cardiology, for example?

Of course. And thank you very much for the invitation. I’m really excited to be here today. As a background, I have a technology background. So I have an engineering degree in computer science, did a PhD in medical imaging electronics, and later on also worked, more the business side and did an MBA at the Valeric school. So I’ve been working for Barco Healthcare for almost 24 years. So a very, very long time, but I’ve taken also many roles in Barco’s organization. I really started as a researcher.

Tom ([05:33].15)

where I worked on the first mammography displays and then did lot of different roles like project management, product management, upstream marketing, quality and regulatory affairs, interesting, patents and IP, merger and acquisitions, some sales. And then the last five to 10 years, I’ve been working, leading a number of ventures inside of Barco that did different things. One of them amongst all the ones in digital pathology.

Currently I’m responsible for the research, innovation and clinical activities of Barco Healthcare. And I’m really privileged because I was able to work in several healthcare domains, including radiology, surgical dental, ophthalmology, and of course also digital pathology. And that really gives me a background and an experience that I now I’m happy to use also in the digital pathology field. Now, you asked me how did you actually start working on co-op technology.

That’s an interesting story. When I started long time ago, my first project that I worked on was actually the first liquid crystal display that was used for mammography applications. And the crucial question that we wanted to answer there was how can we guarantee that mammographers can see tiny subtle signs of early breast cancer in images? So how can we guarantee that even those most subtle signals are visible for a typical observer?

That’s a very interesting question. So when I was presented with that challenge, it required me to deep dive into human visual system perception. So really the theory, the models of how humans, how the eye is perceiving images. So I really did a deep dive in that technology. I also became very active in that research community. It was a really fun period where I had the privilege to work with many talented researchers.

in Barco but also in academics, hospitals, and regulators. And I learned a lot there about how humans perceive medical images and subtle content. Now that was grayscale. Those were grayscale images at that time. Later on, of course, medical images also started to use colors, for example, in surgical imaging and ophthalmology in dermatology.

Tom ([07:56].876)

And that’s when I started focusing also on the color aspects of perception and on color calibration. And the past few years, let’s say the last five, six years, I’ve really been working a lot on color perception, color calibration, color QA, specifically for digital pathology.

Tom, have a question for you. know Alex, it’s your podcast, but I have this question because, you know, men and women perceive color very differently. So you say that you worked a lot on the human perception of color. Have you also studied on the different perception of color for men and women, you know, separately and distinctly on that seeking, you know, which is a food for thought for you.

and

Yeah, it’s a very relevant question, Monica, and thank you for that. There is no such a thing as a standard observer. So every individual, every human individual will slightly perceive images and color in a different way. No, typically those differences are not huge if you’re not color blind, but they are present. So indeed, when we say color calibration and color perception, we always…

refer to a kind of standard observer. what most people are seeing. And you’re right, there are differences between men and women. They are not extremely big, but they are present. That’s right.

Aleks ([09:22].338)

That’s interesting. I’m fascinated by this because this is like, everybody knows what color is, but like, how do you do color science? So that’s what we’re going to dive into. I’m going to start with a couple of questions for Monika. And let’s start with our standard way of examining tissue H &E. And Monika motivated me to put my earrings, my H &E earrings, whoever is watching on YouTube, you will see.

pretty colorful in the H &E colors, pinks and purples. So when we are examining H &E, hematoxylin and eosin stained tissue, what specific color variations are we looking for that could indicate disease? And if you can give us a couple of examples, because that’s going to be crucial to set the stage. Why is it even important to distinguish the colors?

I agree with that because H and E basically forms, I believe, the cornerstone for pathology. And in H and E stains tissue samples, I believe every pathologist knows that color is very important and it gives a lot of information to the pathologist regarding the type of pathology. Now in a typical H and E stain, we have hematoxylin, which we all know is a basic dye, and it stains

the nuclei blue or purple, and whereas eosin, which is an acidic dye, stains the cytoplasm and even the extracellular matrix pink or red. Now, if the cytoplasm stains excessively blue or purple, it could mean an increase in basophilic substances, and this could be due to inflammation or certain cellular damage or even cancers.

We all know as pathologists that in malignant or cancerous tissues, you we generally go by the fact that there is a hyperchromatic, which is a darkly staining nucleus with irregular shapes and increased mitotic figures, as well as abnormal cytoplasmic staining patterns. Now, apart from that, we also have variations in the staining of collagen or other connective tissue disorders or connective tissue conditions like fibrosis.

Monika ([11:36].93)

which could indicate difference in the pattern of staining. And we also see not only for these inflammatory or benign, inflammatory benign or cancerous conditions, we also see a disrupted or uneven staining patterns and let’s say dysplasia or pre-neoplastic conditions wherein we can have variation for staining patterns as well as loss of normal tissue architecture. So I believe

the way we perceive color is very, important as far as diagnosing these conditions are concerned.

I think it’s like one of the main components. there’s, mentioned architecture, you mentioned like specific structures and shapes of things and the color. So it’s like the three things that you look at in a slide, it’s one third of your diagnosis, basically. That’s the color. And I want to go back to what we mentioned at the beginning about the work with clinical trials, because, and we’re going to talk about like how this color

That’s it.

Aleks ([12:40].908)

goes through like a series of processes from staining and all that stuff. But all of these steps like contribute to some kind of potential variability. But in the context of clinical trials, you are going to be dealing with samples from multiple size. And have you experienced situation where the same tissue looked different depending on where and how it was imaged? And not only it looked different, but it like caused problems.

with like consensus between pathologists or in like reaching a specific diagnosis.

You’ve really hit the nail on the head here, Alex, let me tell you that. Because that is something which I think is every pathologist who works in clinical trials will identify with this question or with this kind of a problem they have to encounter while they are working. That’s why we always advocate for a central processing of the slides or images to have more standardized assessment.

Because whenever we talk of a slide or tissue which comes to us, there are certain factors which you cannot control, like let’s say the pre-analytics, which make a huge difference. Simple things like pH of the water can also affect the staining and color. Now have personally seen differences in the color and the intensity of the stain, even within different laboratories of the same organization, while performing staining with validated protocols.

And obviously there is the question of different scanners having different image formats and their ingestion into the whole slide imaging system. So these definitely make the life of a clinical trial pathologist more difficult. sometimes, as you mentioned, that you might have a discrepancy with some other pathologists just because the way the tissue looks or the slide or the image looks.

Monika ([14:37].516)

So it can sometimes form the basis of a study which you can do and do comparative study applying, let’s say, know, different characteristics or formats to certain slides and then compare it in terms of, let’s say, color or color distribution or pixel profile. So that sometimes forms the basis of an internal study as well. But yeah, these things happen. And that is something I think clinical trial pathologists have to deal with all the time.

As you say, color fidelity gives you confidence in diagnosis. Did you have a case where you had to restain the slides or you had to like re-image or something like do some kind of rework with slides because the color was not consistent enough?

Yeah, I think there have been multiple instances, not one instance that I can recollect or let you know, but yeah, because we know that subtle variations in color, especially let’s say with the hematocin and the eosin stain, or even for IHC quantitative tests, let’s say when we are doing an X-scoring where the score is basically identified and defined by the intensity of the color of the…

the stains. I think that is something very important. Even subtle variations in color can be crucial in terms of diagnosing. Yeah. Underdiagnosing or overdrive.

Yeah, we started with my earrings and H &E and obviously IHC is another huge thing where color is basically like even more than 30 % of your diagnosis. It’s basically the whole base for your evaluation, right?

Monika ([16:19].296)

Absolutely. Take an example of something like HER2, you know. A 1 +, a 2 +, a 3 +, is just based on the… is obviously based on, you know, the completion of the characteristic membrane, membrane staining, but it is also based on the intensity of the staining. I think those things make it very, very important that the color matches expected patterns or standards.

And whenever colors don’t match expected patterns, pathologists always feel, let’s say, a decrease in the confidence, or sometimes they find it very challenging. And more importantly, I think this has become important and very pertinent now, because AI algorithms rely on color information to detect features and classification and inconsistent colors across different whole slide imaging systems or different image formats.

can lead to inaccurate feature extraction and reduced algorithm confidence. So I believe this is something which pathologists have to deal and this is something which as we move towards, let’s say, more computational pathology and AI algorithms, the color calibration, color standardization should be more of a norm rather than a choice.

I think so too, and we are not used to this as pathologists because we never had to do it on the microscope. So I think, now that’s going to bring me to a few questions for you, Tom, because I think the education of the digital pathology community about what is color on the monitor, right? And I looked a little bit into your research and it shows that pathology slides can have color gametes extending beyond the standard RGB red.

green and blue, right? And so these are like, this is the standard way of coding color. I mean, there are different options and you’re going to tell us about that, but the pathology slides can have color extending beyond that. Can you explain what it means and why standard computer monitors may fall short when displaying this type of color?

Tom ([18:34].622)

Very good question, Alex. So sRGB is a kind of standard color gamut. So it describes how saturated, how many colors a monitor can generate. And a typical consumer monitor will be able to generate more or less the sRGB gamut. No more, no less than that. So there has indeed been some very interesting research, not only by my own group, but also for example, by Wei-Chun Cheng.

from the Food and Drug Administration. Actually, they published quite recently some results in the Journal of Pathology and Pharmatics. So what they did is they looked at H &E stained images for different organs. I think they looked at bladder, brain, breast, colon, and so on. And what they observed is that the real true colors that are available, that are present on the physical slides can actually be very saturated. So what they did is then they compared

how saturated these colors were and whether or not these could be visualized on a standard display with an sRGB gamut. And surprisingly, and this was also for me quite shocking, they’ve learned that up to sometimes 35 % of the slides contents, so the area of the tissue, so 35 % of the tissue contains colors that go beyond the standard sRGB gamut. So in other words,

You can’t show those colors on a standard display. this, the 35 % was an example for bladder. If you look to other organs, think lung was 10%, kidney was about 5%. And this is only just for H &E images. As Monica already said, IHC images even are much more saturated. So it will even be worse for IHC images. So what does that mean in practice? Well, it means that you can’t visualize those

very saturated colors accurately on a standard sRGB monitor. Not being able to visualize them properly also means that you cannot see subtle differences in color in grading for those very saturated colors. So it’s simply information that is available on the physical slide, on the digital slide, which cannot be visualized on a standard sRGB monitor. You lose data by…

Aleks ([20:55].522)

this course.

Yeah, indeed. It’s clipped, so you lose that information. You have more expensive professional and medical grade displays that have a wider gamut where you can display more colors, but it’s good to be aware of.

Yeah, it’s a little scary that like for some instances you can basically clip 30 % of your data that you use in your diagnostic process. And I mentioned this at the beginning, the ultimate color that you’re going to be visualizing on a display, it can experience different type of variability through the whole

slide production and imaging chain from lab to scanner to display. Where do you see the biggest source of color variation? Like 30 % is a lot. So I would say, oh my goodness at the monitor, but I mean, there is variability across the whole chain. So where is the biggest source?

Yeah, it’s really everywhere and it’s very hard to identify one step as being the crucial source of color variation. It’s really all steps that contribute to the overall color variability. For example, the lab itself, they process the slides and that already introduces color differences. For example, a thicker or a thinner microtome setting will clearly result in a difference of color. Different staining concentrations, Monica mentioned the pH.

Tom ([22:23].225)

of the water that is used, duration you put a slide and a staining, but all of those things will result into color variability. Then you have those slides, you scan them, you put them in a digital slide scanner. Also those scanners introduce color variability. If you take two different brands of scanners, they will generate, even if you scan the same slide, two differently looking images.

But even if you take the same brand of scanner, same type of scanner, you will experience sliders, but still present color differences between individual scanners. And these can even change over time. Moving down the imaging chain, then you have the viewing software. Viewing software can also have an influence. In principle, you could think, well, the viewer should just perfectly render the content of the file. Yes, that’s right. But that’s also theory because

Very often those viewers support a lot of different file formats, often proprietary file formats of different scanner brands. And then the way they read those images, those image files that can be slight inaccuracies, again, resulting into color differences. And then finally, there is a display system. In the end, the display is what the pathologist is looking at. So this is also very important component.

There is a lot of different displays people can choose from nowadays, going from cheap consumer displays over professional and medical grade displays. And of course, there is a big difference also in quality and in characteristics of those displays. So all of them really introduce variability. If you ask me the question, where is the biggest source of variability? Well, I think the lab still today contributes

to most of the variability, but it’s not going to be 50%. It’s going to be rather 30, 40 % of the total color variability. So you really need to tackle all of them if you want to have a end-to-end color consistency.

Monika ([24:28].266)

You know what, Tom, I completely agree with you when you say that these are the steps where you can have difference in color consistencies, whether it is the staining protocols with a particular laboratory, the whole slide image scanners, because every scanner, could have a different image format. And even different scanners from the same company, depending upon the model you are using, can have different formats.

And then obviously the monitors can make a difference. Apart from these three crucial steps, I think there is another important step, which is the pre-analytics. You know, how the tissue comes to the laboratory in terms of how it is well preserved, how it has, you know, what is the time taken to reach the laboratory? So all those things also make a difference to the color. If you see a tissue which is very necrotic, it starts appearing too much of a pink. So I think those things also make a very…

important contribution. But unfortunately, not all the time the pre-analytics is under your control. You can only do as much as what happens in your lab.

It’s a very multi-step process. like how I see it is like each of these steps contributes and we cannot neglect any of them. So, so to me, like the color aspect is like, it’s, it’s at the end, we already dealt with so much, like we kind of tend not to pay enough importance on something that we can actually control. Whereas the…

previous steps of the imaging chain of the slate production chain. Okay. If you’re like doing clinical trial work, what are you going to do? Nothing. You’re going to get what you’re going to get. So you make the most and the most consistent samples for the pathologist out of what you get. Right. And obviously not forgetting about what went into producing these samples.

Aleks ([26:26].144)

Monika, I want to ask you a question that is kind of triggering for the pathology community because it’s about speed of evaluation. And why is it triggering? And it’s triggering every time you say speed for a pathologist. Me personally, feel like don’t make me work faster. I want to work, I want to use the amount of time I need to use to be confident in my diagnosis. I don’t care how much faster the software is. I don’t care like how much faster.

digital is compared to what glass would be. I want to take the time I need to become confident and comfortable with the report I’m signing with the diagnosis. Right? So, there’s this speed trigger factor. But what I want to ask is that the color research, Tom, your research shows that the read can be faster on calibrated display by a factor of 28%.

In your experience, and not only according to color, but in general, what makes the difference in reading speed and how much like your confidence in color contributes to you being as fast as your expertise allows you to be.

There are two questions here in this particular question which you’ve asked me. I think one, I would like to just focus on the fact that in my experience, working on a digital display makes me much more efficient than manipulating this light under a microscope where my neck

I’m in the same cab.

Tom ([28:03].49)

Totally.

So I totally am a believer and also, know, over the years I have realized that navigating a sample, sorry, navigating a slide, especially when it comes to resection samples becomes much more easier with digital display monitors or digital display systems. Also you have with digital displays and monitors, you know, you have features like synchronized pan and zoom, which enables quicker navigation and analysis.

Also, digital display systems facilitate side-by-side comparison of images, which can be very useful to look at immunohistochemical analysis. You are looking at a particular cell on the slide or on the image, and then you try to just identify and make sure that it is exactly the same cell you are looking at the next section. So I think those things make it much faster, easier. Also, ergonomically, I believe, you your

your necks and your shoulders are not under so much stress when you are looking at a digital display system. And that is something which I think we pathologists feel that that is important. And also, it’s important that digital platforms also facilitate side-by-side comparison of images. You are looking at a cell.

in the HND and you want to look at the same cell in the IHC slide and then you can compare a side-by-side image. Much easier to be done with a digital display system than under a microscope where your neck and your shoulder is always under stress. So I think that way it becomes much easier to navigate the slide also and it is less stressful and very more time efficient I would say.

Monika ([29:54].392)

Now, as far as your next part of the question is concerned regarding calibrated displays, I think calibrated displays, what you get is a consistent and reliable color reproduction of tissue samples, which is important and essential for identifying subtle features. And importantly, not only the color, the brightness also is very consistent and reliable, leading to,

proper interpretation of the structure, the diagnosis, and the slide. So I think calibrated displays make it much more efficient in terms of time because you are seeing a very reliable and very, I would say, light and also reproduction of color. So I believe that makes it much more efficient for the pathologist. Also, it’s easy on the eye. That’s something I have noticed in my experience.

If I can make a comment on this one. Yeah. So it’s not only our research that has shown an increasing reading speed of independent other groups that came tomorrow as the same conclusions. Now, when we anecdotically spoke with pathologists that were involved in the clinical trial we’ve done for this, there are actually two things that repeatedly are being mentioned. So the first one is about confidence in the diagnosis.

If you have a high quality, reliable display system, you feel confident that what you see is accurate and you feel confident that what you see in these images is actually what is present in the digital slide image. And that just makes you go faster. If you’re not confident in your visualization.

Aleks ([31:40].482)

you the decision faster.

Right, you won’t trust it and you will go back and you will double check and you lose time. You simply lose time. So that’s the first thing that we got as feedback. The second thing we got as feedback and that has to do with what Monica mentioned already, specifically for grading tasks. If you properly calibrate your display, it’s easier to see different color tints and also color tints that are not that different from each other. So when you need to do a grading task.

That’s exactly what you need to do. You need to judge what kind of color tint do I see it, which one is it. So on a calibrated display, that’s easier. In other words, you can do it faster. Maybe Monica can give some real life experience or some real life comments about that.

Yeah, I can definitely add to that. Another thing, you know, apart from the color, the light also makes a lot of difference. You know, you have very, I would say very uniform appearance of light all over the tissue. So even some areas where you know that you have not, they are not fixed properly or they are not stained properly. think you are able to interpret better, especially when you have better light on the tissue. Also with calibrated displays I have seen is that the over time there is

less deterioration as far as seeing the tissue or slide is concerned. With non-calibrated monitors over a period of time, you go to them, then possibly there might be some degradation the way you see a tissue or perceive a slide.

Aleks ([33:10].72)

Yeah, that brings me to my next question. just to finish that speed question, it’s basically like with quality calibrated displays, you have all the information available to you immediately. You don’t have to like hunt for them in different area of the slide, right? And if you have all the information that you need for your diagnosis right there, you can make your decision faster and that contributes.

to the speed because you have all quality and information necessary for diagnosis there. And now I want to follow up on what you started talking about, Monika, but my question is to Tom about the degradation problem. In non-calibrated displays over time, there is a dramatic color shift. How? Like I have never seen a dramatic color shift on my monitor. Can you walk us through what happens in the display?

over the lifetime without calibration. Because obviously it’s not something that happens in minutes, it happens over time and when something happens slowly we don’t notice it. But how does this influence the work?

That’s a very good question. Let me answer that one with a little more technical details about displays. So display systems, they really consist of multiple optical components. For example, liquid crystal displays, which are the most commonly used type of displays still today, they consist of a backlight with light-emitting diodes, LEDs. And these generate light at the back of the display. And then you have a liquid crystal layer that modulates

that light on the backside and generates the image. Now, if you look at the backlight, so those individual LEDs, when they’re brand new, they have an initial color point and an initial efficiency. Over time, the efficiency of those LEDs will degrade. That’s the same like when you have an LED light bulb in your house. In the beginning, it will give more light or it will emit more light than after one year or two years.

Tom ([35:19].35)

And also the color will shift slightly. So with those backlights is the same after six months, 12 months, 18 months, the brightness reduces and there is also a shift in color points. On top of that, there are also a lot of other optical components and I will not go into the details, but we have collimation files, have polarizers, you have all kinds of optical components that typically deteriorate under UV.

emission, UV light that is actually deteriorating those optical components. the LEDs got used up, yes. And the other components just degrade and change in transmittance and in color point. But it’s even worse than that. Because like, for example, the LEDs really fluctuate in function of temperature. So on a really hot day in a hot office.

They basically get to use…

Tom ([36:15].202)

You will have a noticeable difference in efficiency and color point of LEDs than when you’re in an air conditioned room with always the same exact ambient temperature. So those are the reasons why display systems that are not calibrated change over time. And you’re right, Alex, you don’t notice that if you just have one display. If you buy a display system and you sit behind it for two years, this is a gradual shift and you will not notice.

But if you would put a brand new next to one that is already six months old or 12 months old, you will immediately see the difference. And that’s a reality in a reading room where you have typically multiple desks, multiple workstations with displays that have a different number of hours being used. Some of them are newer than other ones. So you will really see differences in both luminance and in color appearance between those displays.

also believe, you know, Alex, not only with the time degradation, there is also difference in the uniformity of the light, which a pathologist perceives over a period of time. And that I have seen really makes a difference in the speed with which you evaluate a slide and the confidence in which you have by giving out a diagnosis or assessing a percentage for any of those immunohistochemical biomarkers. But what is very interesting, Tom, from your

which I didn’t know about is the temperature of the rain can…

I’m immediately going to the image of somebody reading on the beach. And now not happening anytime soon.

Monika ([37:53].568)

Exactly. And not only on the beach, you somebody who’s sitting in, let’s say in Edinburgh, you know, and seeing a slide versus somebody who’s sitting in a tropical country and seeing it on an uncalibrated monitor, could, it could make a difference.

And especially to come back on the explanation that I gave, the differential aging as technical people would say is a problem. So it means if you, for example, buy 20 brand new consumer displays, they more or less look the same when they’re new. But after six months, they will start looking different because one of them may be used more intensively than another one, or there is slightly difference in aging characteristics. And that results in a situation which is really…

I think unacceptable because as a pathologist, you will be faced with situation where depending on what display you use, depending on what workstation you pick on that day, you will have a different appearance of the slides.

And all of them would be validated stations, right? Regardless. Yeah, I think it’s it’s an important factor because many labs, well, color is important. That’s why we’re talking, right? But what I want to emphasize here is that the decision which monitor to choose is often based on the available budget. And yeah, maybe being guided by the price of it is not the best way to choose these devices.

It’s, I personally, think it’s very short sighted to purely base the decision of what monitor to purchase on the budget. Yes, financially it looks more attractive at first sight, but it actually is not over time. Because those consumer displays that are cheaper, they will not last for, for example, five years. You will need to replace them sooner because they were worn out after, I don’t know, one year, a year and a half, two years. Well, typically.

Tom ([39:46].528)

If you buy medical, high quality medical displays, you have a warranty of three to five years. So for every medical monitor, you’ll need to buy two or three consumer displays during the lifetime of that medical display. But there are also other aspects that you need to take into account. Confidence of the pathologists, we already mentioned, higher reading speeds. I mean, if you start making the calculation, investing…

Mm-hmm.

in a high quality monitor, but gaining 20, 30 % in reading time, I think it’s an easy calculation to see that you actually have a positive ROI to invest in higher quality equipment. If you have a multi-site lab, where you’re reading at multiple sites, where they may have different space systems at multiple sites, I’ve seen it several times where pathologists from one side is discussing with a pathologist in another site and they don’t agree.

on a slide and then they start talking and they say, I see this. And then the other pathologist says, yes, I don’t see that. And then in the end, you dig into that case, it often comes down to core variability and very often the monitor. But the time you lose with that, and the time and the cost that you need to spend to reprocess slides, you.

you think there something wrong is also a cost. again, I think it’s quite short sighted to purely decide based on the price, need to look at cost of ownership and beyond that the total picture of what a high quality display can bring.

Aleks ([41:15].278)

Mm.

Aleks ([41:27].79)

So the problem that I see in this being like the last step of the imaging chain and like basically that is unfortunately the place where people will want to cut corners where they should not cut corners. Basically just by the whole, just by, because of the fact that this is the very last step and you already have invested so much and you already have done so much work. So yeah, we cannot do that.

I just wanted to highlight that I think for any hospital or an organization to make a decision regarding investment into calibrated monitors or into calibrated display systems, I think it’s important to process it more in terms of time, let’s say, because the increased turnaround time, this word is so effective. And so I would say it…

gives a very good response from the business side of things. So when you say you have increased turn-around time means that the pathologist, if that pathologist has seen 20 or 30 slides a day, that means that pathologist is able to get more slides out that day. And this particular efficiency really appeals to the business side of things. Another way I have realized is when you are looking at different display systems and

also deciding on which monitors would be the best for you is to do a little internal study wherein you can compare different monitors and, you know, with different pathologists. And then you probably have quantitative and qualitative data to go to your business team to make them realize that, okay, maybe this is looking a little bit more expensive upfront, but possibly over a long period of time, it will lead to better efficiency and decreased costs.

Knowing where to focus your effort on the due diligence of the whole imaging process is super important. I want to focus a little bit on technology here because the, like, let’s say the three most connected pieces of the imaging chain will be scanner, the viewing software and the display. So Tom, I wanted to ask you first about…

Aleks ([43:45].186)

How can we decrease the variability of coming from the scanner? And here I’m referring to the international corollary consortium profiles. And then what does calibration look like in practice? Like how do you even do it?

Yeah, it’s very good question. indeed, so ICC profiles from the International Color Consortium is something which has been used already a long, time in the, for example, the prepress industry, but which is quite novel in the digital pathology space. So ICC profiles, are essentially data files that describe color characteristics of a particular device, like a monitor or a scanner or a viewer.

So they really describe how, for example, a scanner will translate real physical colors that are present in the tissue on a slide into pixel data in RGB values. red, and blue pixel values that are present in the slide. So each scanner, as I explained already, is going to do that in a slightly different way. If you look to completely different band of scanners, it may even be very different way of doing.

So the problem is that if, example, as a viewer, you do not have information on how that scanner behaves, or how the scanner does the translation from real colors to RGB values, you also don’t know how you need to visualize that slide to make sure that if you look on the screen, it really looks like the real colors on the slide because you lack information. So if you have ICC profiles of both the scanner and the display,

Then you can put those together and as a viewer you know, okay, how did my scanner do the translation? What will my display do with RGB pixel values? So how will the display actually translate those pixel values into light, into colors? And the viewer can make sure that all of this is done properly. So you need an ICC profile on the input side, on the output side, and then you really can have an end to end consistency of the colors.

Monika ([45:27].416)

Do it, Camila.

Aleks ([45:54].006)

And where does it come from? Like does it come from the scanner? Each device comes with it? Like practically, how do you get hold of it and introduce it into your imaging chain? Because I’ve been, I’m on the receiving end of viewing these images and I can like click on the ICC profile in my viewing software, but it’s not always there. Whenever it’s there, I see the benefit because I don’t have to like manually move the sliders and

adjust the image to the way I want to look at this image to like have the optimal contrast, which I am not consistent with myself. It’s going to be different from like one study to the other study, but sometimes it’s there and sometimes it’s not there. How do we make sure it’s there and we can benefit from it?

Really the ICC profile needs to be present for all components in the editing chain. So it needs to be there for the scanner, viewer needs to support it and the display needs to also provide an ICC profile.

So on the device level basically.

It’s on the device level indeed. And it’s not really something you can directly influence. only way to influence it is to put pressure on your vendors, scanner and display to actually make sure they support ICC profiles and they support an accurate ICC profile with the devices that you purchase. And when I say an accurate ICC profile, ideally I mean an ICC profile, which is really specific for that individual

Tom ([47:22].702)

scanner or display because some scanners vendors and some display vendors provide generic ICC profiles. So they, for example, provide a profile which is the same for every display of a certain type, but it’s not optimized or not individual for one particular display. So if you really get an individual ICC profile, this will give you the most accurate

color management and most accurate color visualization.

I think it’s important to highlight that what you mentioned that it’s on us on like this way of influencing things by like putting pressure on the vendors is a recurring theme in digital pathology as a pretty new area of imaging diagnostics, I would say. And it’s still a very dynamic process where it can take place where when the users are

informed enough, educated enough about the process and not only, you know, their area of expertise, not only pathology for pathologists, but enough information about what goes into the end product. This is where you can have this discussion where the multidisciplinary aspect of digital pathology can shine and where you can influence different aspects of the workflow to have the best.

product at the end that enables the most confident diagnosis.

Tom ([48:57].002)

Absolutely. this podcast is very important in that context because it’s about market education to a large extent. If end users, if pathologists, if medical physicists are not aware of this, then there will be no pressure on the vendors, on the manufacturers to actually make sure that we can achieve this. Because if not all, again, if not all the components support it, it doesn’t really work well. So I’m really happy that I’m here today and that I can be part of that advocacy towards.

the antivirus.

Yeah, this is important. This is super important also in terms of training, like we said, internal teams. And here’s my question to you, Monika. You’ve been in the digital pathology space for quite some time and I am sure you’ve onboarded a lot of people who now are digital pathology trailblazers. When you train new team members on digital pathology, how do you approach the aspect of color?

What are the information that people need to be aware of?

I think it’s important that whenever you are onboarding people who are working on digital pathology or who are new to the system, as already Tom mentioned, that we need to have a consistent color space throughout the entire workflow, know, from let’s say image acquisition to display. So, I mean, I would advise or I would at least suggest that, okay, you calibrate each monitor, create ICC profiles.

Monika ([50:27].252)

Use a consistent color space throughout the entire workflow from image acquisition to display. Then you also need to train team members on how to visually assess color accuracy and look at inconsistencies. This can be done by providing phantom slides, which can be used to assess color reproduction across different workstations. And also, it’s important to…

to apply or to implement a comprehensive color management strategy. The training should also emphasize the importance of color consistency and provide practical guidance on how to identify and address potential color variations. So whenever we have an issue where in a pathology laboratory has really, they are just entering the field of digital pathology and then we try to

advise them or try to navigate their journey into this field of digital pathology by identifying issues wherein these color differences can actually be caused. so these are, in my opinion, we talked about the pre-analytics, but we need to say that there is inconsistent preparation process of tissue to glass slides, sometimes with non-standardized staining protocols.

Then there is inconsistent optical and hardware characteristics of the scanners and different color image formats. And then obviously we have image monitors which are different, which already Tom has discussed. So how do we take these things into consideration? The first is obviously we can do a stain normalization for standardizing, let’s say stain color or appearance in histopathology with slight imaging without affecting or intervening

intervening too much in the pre-analytical phase. The second thing is an internal color calibration, which I think most pathology laboratories have now started to use, wherein we are using a calibration slide with a known truth of colors. And this color calibration slide, as you know, it’s available commercially also and with different color patches on it. And you can also make it yourself in the laboratory.

Aleks ([52:48].686)

It’s basically like with squares, right? With color squares that have defined.

It has like what you say, know, these number of color patches on it together with calibrated colors or like transmission spectrum, which can be made in-house. And right now we have a lot of commercial vendors also who are doing this. so that is something which you can do as for internal color calibration. And obviously we have spoken about external monitor calibration using a

colorimeter or spectrophotometer for aligning the appearance of colors. But this is obviously in an ideal setting. At this moment, when we know that there are many pathology laboratories who just started getting their first steps into this world of digital pathology, I think these changes are definitely recommended, but will take some time to be actually put into practical application.

I think like Tom, you said that it’s super important to learn about this because I think it was a little bit of a neglected aspect of digital pathology at the beginning because everybody was focused on the scanners and you know, scanner still is a central component to the digitization process, but it’s not just the scanner. before we dive.

So obviously we need to talk about AI, we already started, but before we dive into AI, I wanted to ask you about, so you already mentioned this term and color management processes for color management, which is, want to emphasize as well, but also in regulated environments like drug development, obviously primary diagnosis, and how do you validate?

Aleks ([54:38].382)

color reproduction, how do you make sure that it’s consistent and accurate? Like a practical way of doing it or like who takes care of it? Like where does it fit into the validation process?

I think that’s a very, very pertinent question, you know, because the transition to whole slide imaging systems has been gradual over different organizations, different countries. And you also have key regulatory bodies like the FDA, which now require better color fidelity alongside other things which determine

diagnostic reliability. FDA now wants a separate module for analysis software in the medical approval process. and that analysis software module is required by the FDA for the medical approval process. And that can only be achieved if colors in the image are being analyzed at a particular standard. how do you do this? know, what FDA recommends is the use of a target slide.

Now this target slide should have like a set of measurable and representative color patches. So ideally the colored patches should have similar spectral characteristics to a stained tissue. The FDA also recommends to create three data sets, injected color, and output color. So essentially this means that the measure of a true color of a stained sample and comparing it

with the output of a whole slide device to identify the device’s effect on color representation throughout the digitization process. So I believe creating a truth data set is very, very critical to the whole process and requires the use of specialist equipment calibrated to internationally accepted standards, like how Tom was mentioning, depending upon the ICC requirements or the ISO requirements or standards.

Monika ([56:45].218)

Now, this is something which is definitely recommended, but I believe there is still going to be some time before this is. And also we actually implemented, but obviously we have, we’ve seen in the recent past, you know, in April, we saw FDA clearing the first use QCS algorithm to be implemented in clinical trials prospectively. So we are definitely taking steps towards.

I believe so.

Monika ([57:13].838)

getting more stringent regulatory advice on how this will be implemented in a real world scenario. So it is going to take some time, but I believe as more and more integration of AI systems and AI computational algorithms come into this space, I think this particular field will have more regulatory advice and more stringent regulations, not only about other systems,

and other aspects of digital pathology, but also color. that’s my understanding.

Absolutely, Monika. maybe to complement what you’ve just said, it sounds very complex doing all those processes and executing that management process.

and costly and time.

Time consuming, requiring lot of expertise and trained people.

Monika ([58:10].656)

convincing people to actually do it and go through it.

Yeah. Yeah. But some of the vendors are really going that way and are supporting it. At least for the displays, if you buy a high quality medical display, it typically comes with software that will automate that kind of tests for you. So as an end user for sure, but also as a responsible for quality of a lab, this is not something you still should do manually.

It’s something that that software can do automatically. The display is full of sensors, can do the measurements even without you knowing, will generate a report and then inform you if there is a problem. So that push from FDA, rightful push from FDA towards more quality, I think will also have an influence on the different vendors and convince those vendors to actually go with the flow and to make sure that there is more attention and more support for proper color calibration.

That’s good to know. I mean, in digital pathology in general, I think that collaboration between users and vendors is pretty close and we do rely a lot on the vendor support for this types of things, especially when validations are considered and anything in the regulated environment. And Monica, you mentioned algorithm approval. So let’s dive into the AI aspect of things. I’m going to start with.

on this one, because the main shift in terms of color and in terms of developing algorithms for generalizable AI algorithms that will perform on different scanners is a big discussion. Is this even possible? So let’s start with how does the color variability impact AI performance?

Aleks ([01:00:10].222)

And second part of this question is like, how do you deal with it? Like, obviously you may have plenty of different scanners in, well, plenty. There are only a limited number of scanners, but you can have different scanners in the same institution where you want to develop algorithms. Tell me your opinion on that, different color from different scanners and AI algorithm training.

Yeah, it’s a very hot topic and it’s also a very important topic with adoption of clinical AI being more and more widespread. Let me take a step back and tell a little bit about AI algorithms. So the way AI algorithms are developed is you start with collecting a training data set. So you collect a large set of images with ground roots. Ground root could be, for example, the diagnosis, could be grading, could be anything you want the AI algorithm to predict.

you present a training set to AI algorithm to get the AI algorithm and it will learn from those examples. Now it can only learn what is present in the training set, which means that if you feed the AI algorithm with images that are all taken with one type of scanner, it only sees those examples during training phase. It will learn from it. And when then you present, in the inference stage.

another image to that algorithm that comes from another scanner, it has not seen that before. And actually it will have a much lower accuracy compared to when you would present it with a scanner image that has been trained for. To quantify that a little bit, there is plenty of scientific research on that. Your accuracy can easily drop with 20-30 % if you present it with an image which it has not been trained on. So a scanner image.

It has not been trained on. that’s huge. It means that can make the difference between a clinical algorithm that is really useful, really accurate, and something that is just not useful in practice and not even safe to use. Now, how can you overcome that problem? How can you solve that problem? I think there are three main ways of doing it. The first way of doing it, and this is what many clinical AI algorithm companies has done so far, is they limit scope.

Tom ([01:02:28].994)

They say I’m going to develop an algorithm and I only support this type of scanner. And I create training data for that type of scanner. I do my validation on that type of scanner. I go to FDA and I got a clearance or an approval, which limits use to that type of scanner. That’s handy. It’s clear to test, to validate, and it also limits the investment you need to do, but also your scope and also your commercial usefulness.

is much smaller, but that’s what typically has been done so far. And the second option is to say, no, I do want a broad applicability of my algorithms. So I’m going to collect a training data set, which is really representative for all types of images that regular users would encounter and would want to the algorithm on. Yeah, that can work, but it’s much more expensive.

You will need to collect a lot more images. will need to include much more sites in your clinical trial. And also your validation will be much more expensive. But if you succeed and technically this is perfectly feasible, you will have an algorithm which is broadly applicable. And if you’ve done everything properly, which will achieve regulatory clients. But again, more time, more investment. not something that many startup companies…

prefer because it takes them also much longer time to market. And then there is the third option. And that brings us back to the discussion we had in this podcast already a few times. If you can standardize your color, so if you could for example make use of ICC profiles, if you could use calibration slides, and you have a way to actually standardize your color, normalize your color, make sure that any input image from any scanner can be mapped

up on the standard normalized format. if you then you train your AI algorithm on those normalized images, it’s much easier because you do not need such a large data sets. Your AI algorithm also is presented with images with less variability. So it will train faster. Your training process will go faster and your applicability is much broader because as soon as you present it with a standardized image,

Tom ([01:04:46].508)

doesn’t matter what scanner it comes from, as long as it’s standardized, the performance, the accuracy should be acceptable, should be okay.

That sounds like the perfect solution.

It’s indeed, think personally the future way of working. It’s something that is not common today. It’s something that I’m not aware of any company which really has done this process completely. There are experiments, certain academic experiments in that way, but I’m truly convinced that this is the future.

Yeah, logically, it just sounds so simple because like you just map to what it’s supposed to be and then train on that. Yeah.

Also, I would like to add, think, let’s say four or five years earlier, we were, as Tom mentioned, that companies or vendors were looking to create algorithms only with a particular scanner or a particular image format. Now, what was happening was that this was making their use very, very restrictive. Only if you have a particular scanner, a particular image format, you would be able to access that algorithm.

Monika ([01:06:01].94)

Now, over the last two or three years, I’ve seen a trend with different vendors wherein they are trying to have more extensive datasets coming from different scanners. But till we have what we call a very standardized imaging system, something like we are talking about DICOM right now, I think that is a persistent problem which we are all going to have over different laboratories or different organizations.

So that is something I think we need to be very cognizant of. And also, using medical grade displays with a wide color gamut and accurate color reproduction. I think that is important, especially when you’re using AI algorithms, because AI algorithms rely on color or intensity information to analyze tissue samples. Now, we’ve had so many of these algorithms which have actually come out only from H &E slide.

And when you know and we talk of H &E, we know that it is all an interplay of beautiful colors, whether it is pink or whether it is blue or whether it is purple. As evidenced by your…

My earrings? Monica, do you use any algorithms? Do you use any image analysis in your current work where it’s crucial? I do. do. What do you use

…analysis algorithms for, because I work on ADC therapeutics and especially antibody drug conjugates. So we definitely use algorithms for analyzing these different antibody drug conjugates, not only for their, you know, their appearance, but also looking at their spatial characteristics in response to immune cells and in response to other cells in the vicinity and in the tumor microenvironment. We are definitely doing that.

Aleks ([01:07:49].038)

Yeah, and IHC as we already said, it’s crucial because this is where color is what makes the diagnosis. getting a little bit more practical, we kind of talked about the return on investment question where basically what I’m hearing and add whatever you want to add here, both of you, but what I’m hearing is that

short sightedness and short term cutting costs is going to haunt us in the long term. We said that like in the lifetime of a medical grade display, you will probably replace a consumer display two or three times, right? The speed of diagnosis. So what would you say if the decision makers would still see

the investment into color calibration and monitors as an optional expense.

You know, I always, because whenever we are wanting to convince people, you know, who make those decisions to buy a particular, let’s say a particular monitor or a particular scanner, I think the word, the increased start time, the turnaround time, very, very effective as a business decision.

I’m in ya

Monika ([01:09:18].05)

And also optimal quality of evaluation because both these things in terms of, you know, business profit profitability, increased turnaround time means your pathologist hours are much more valuable. You are able to do more in a day. That means you have better clients, you have more clients. That is one thing. Your turnaround time, think, whether in the hospital or in the industry is very, very important. And the second is obviously optimal evaluation.

And also, I think, you know, in terms of reduced errors, you can also quantify both because time is money. So you are also quantifying time as money in terms of reduced errors and in terms of reduced, you know, second opinion, consultations, delays. So I think all those things put together, that makes a good business case. And as we discussed in the podcast earlier, and I alluded to a study also, which we did.

internally where we compare to different calibrated displays and different monitors to see in terms of not only quantitative but also qualitative difference these calibrated monitors could make in terms of diagnosis, in terms of evaluation, and also in terms of readability. So I think those things are very, very crucial whenever we are making a business case across any organization.

So Tom, let’s say an organization says, okay, we acknowledge the importance of color, but of course we already have all this other equipment. How would you advise, or like, how should they tackle the color calibration and let’s say color quality in general to like put everything together at an organizational level? Is it something they should do like replace everything all at once or gradually introduce

the new equipment that is then compliant with regulatory guidance, how would you guide an organization to take color seriously, basically?

Tom ([01:11:24].556)

For new deployments, course, I can only stress that it would be not wise to ignore color consistency because you’re going to invest millions in infrastructure that you hopefully will use for at least a number of years, if not longer. Things are moving so fast, so much more insights about the importance of color, regulatory changes that are happening. So it would not be wise if you invest today, not to make sure your entire imaging chain

is properly color managed. Now things are different if you already did the investment. And then I don’t think it’s feasible or I think it’s to be expected that people will replace everything. Now that’s a matter of indeed step-by-step replacing components. This is technically possible. It will not be perfect, but it’s going to be, it’s technically possible. And with every step you take, you will move towards more color consistency.

Now, if I may go back to Monica’s last answer on how can you convince organizations, I think there are two more things that I would like to add from my side. Ergonomics is important. Pathologists spend hours and hours a day looking at the monitor. I think they have a right to use a monitor that is good for their eyes, that gives them a good ergonomy. So that’s a very important one.

yeah, definitely.

Tom ([01:12:50].558)

And the second one, and this is one that most organizations don’t think about, many organizations also store their data with an intention to monetize the data. So at some point they try to sell the data or do clinical trial services or whatever. More and more it’s the case that.

people that want to buy data sets, but also regulatory authorities that need to review drug applications or other medical device filings, they actually are going to ask you questions on how you did your color management, the quality management of all your components. So what I’m trying to say here is if you collect data and you completely ignore color management, and then you try to sell your data later on, one of the first questions you will get is what about your quality system?

Can you prove to me that your data is high quality? And if you don’t have to say, ignored color management and I don’t know what monitors we read on and how we created the ground truth, the diagnosis, I can tell you, you will get a lot less for that data.

That is such an important point that I didn’t think about basically in the time where data is the new currency. Like you have to invest in the quality of your data and this is part of the quality of your data. Interesting. Huh. That was a aha moment for me. Monika, we already talked about the regulatory aspects of this and validation, but I want to add one question here.

What is the role of professional societies in establishing color? I would say even not, not only standards, but basically like color literacy for digital pathology. I don’t see it being part of any kind of education. see a lot of, how are images generated in a scanner, how to validate the scanner, all this like very clinical focus. I don’t see enough about.

Aleks ([01:14:57].934)

Do you see this popping up now or what’s your opinion on that and the role of societies in that?

I think because, you know, I am a part of the Digital Pathology Association as an Education Committee member. So I know at this moment that there are certain, I would say, discussions regarding the color displays and color monitors and how this is very, very important as far as digital pathology evaluation is concerned. So I think there is more and more discussion which is happening not only in societies, but also, you know,

among the digital pathology community that this is an important aspect and this needs to have more, I would say, regulatory guidance and advice as to how organizations are able to implement this in their digital workflow. So I believe there is going to be more discussion around it. And I think as we know, you know, that each of these discussions or decisions are then precipitated like

I mean, I give you an example that for the use of digital pathology systems, we knew that already there was people were interested in it and there were organizations who were adopting this. come pandemic and then it really precipitated the need for digital pathology because people were not able to go to the hospitals or their organizations to work.

led to that, you know, the decision that, okay, you could remotely review the slides and then evaluate them from the comfort of your home. And there was a big CPS as well as FDA decision around that point of time. So similarly now, two months earlier, when we say that, you know, there is a FDA recommendation, FDA approval for the QCS algorithm to move to the clinical trial phase, I think these are steps which

Monika ([01:16:58].146)

the industry as well as the digital pathology community and other QOLs have started already taking regarding having a more, I would say a more regulatory advice regarding not only a digital pathology, but also color calibrated systems. So I think each decision in the industry basically leads to more movement in the digital pathology community and industry as a whole.

And every time there is some sort of a major push in terms of, the regulatory climate or in terms of other conditions like pandemic, we’ve moved forward. And I see this as one of those points wherein there will be more regulatory advice and more stringent regulations around digital pathology, not only for colors, and especially now when we see that we are looking to incorporate AI algorithms

as an adjunct to the pathologist. Now that is a very, very important aspect wherein I’m sure the regulatory bodies are looking into it where you are using AI algorithms as an adjunct to make sure that the color displays and the colors from which the AI algorithm data set and the ground truth was devised is very standardized across different laboratories in different countries. So that’s my take on it.

Yeah, think guidance is a keyword that I’m hearing here. Guidance for the users, but the guidance is going to be created by societies interacting with regulators and establishing common understanding, the minimal requirements and then communicating it to the users so that we don’t have to reinvent the wheel. I think that’s going to accelerate the proper practice of digital pathology.

Before we close, I want to ask you about how do you think the future is going to look like? So Tom, I’m going to start with you. Anything new happening in the industry right now? Any new display technology, something that’s going to like be good and you don’t have to touch it and it’s going to be fantastic? like, where do you see this field going in like five to 10 years?

Tom ([01:19:20].238)

Let me start with answering the question on the display side. I think there is still room certainly to improve display systems optimized for pathology. So today you already have very high quality systems, but with new stains being used and so on, you actually have room to further improve your display as well related to resolution. Refresh rate is a very important one. Co-orgam it. We talked a lot about.

So displays with more broad color gamut. So then you immediately go into the direction of, for example, all that displays that could replace the more traditional liquid crystal displays that are mostly being used today. So on the display side, room for more innovation, for more technological advances. But the other thing that I want to really strengthen or mention is Daikon.

So DICOM standardization and particularly DICOM working group 26. So DICOM working group 26 is a group that is specifically focusing on working out the DICOM standards, which is used already for more than 15 years in radiology and make that really applicable and tailored for digital pathology applications. They’re not only looking at interoperability file formats and so on, but they specifically are also addressing

topic of quality control, core consistency, ICC profiles, and so on. And this is a very, very active research group that is meeting many conference calls and also physical meetings at all of the major conferences. And I really expect them to bring out important guidance in the years to come. And that’s also a recommendation I want to give to labs or sites that are now going digital, that are going to invest.

into a digitization project, I really recommend that they adapt the DICOM standard and not use proprietary image formats, for example, suggested by vendors. Because really DICOM will become the standard and DICOM will support in an easy way many of the quality control consistency aspects that we discussed today.

Aleks ([01:21:38].274)

Yeah, we could not have a Digital Pathology Podcast episode without mentioning Daikom.

Thank you.

Yeah, I mean, this is the standard that we have and there is still work to be done to implement it across organizations. So Tom, if you could implement one change across all digital pathology systems tomorrow to improve color accuracy, what would it be? Like one thing.

If I can only pick one thing, then I would say start using ICC profiles. That’s the first important step everybody needs to take. And using ICC profiles also means make sure your device is supported.

Monica, same question to you. If you would tell a pathologist who’s skeptical about the color calibration, why should they care? One thing that is most important.

Monika ([01:22:30].072)

You know what, to be very honest, think any pathologist who has gone

They’re at the care, right? Yeah.

rigorous training procedures which we go through for so many years of our life, they know the importance of color in pathology. It is not just about pink and blue in H &E, but also the beautiful different stains and colors which you see across different, you know, different stains, be it let’s say, past staining glycogen or mucin, a bright magenta, or metachromatic stains like a gene sour or ptolidin blue. So I think that these colors really produce

Mmm.

Monika ([01:23:06].284)

valuable diagnostic information and helps the pathologist to identify disease and to determine the effectiveness of treatment. So I would personally say that at least half, I know sometimes things can get very expensive by the time you, as you mentioned, this is the last step of the procedure. So you’ve already invested so much money into the scanner. You’ve already trying to connect your laboratory information system and other IT.

IT support and infrastructure to all of this and at least try to have a color calibration slide and if you can go for a color calibrated monitor to make sure there is a formity of colors across the systems.

Thank you so much to both of you for joining me today. There’s a lot more to this topic, but I think we kind of managed to create a basic understanding why it is important and about both the technological and pathology aspects of why people should care. What I’m going to do in the show notes, so whoever wants to dive deeper, Tom, I’m going to include all the…

publications and all the research that you mentioned. And there are a couple of resources provided by Barco, who’s the sponsor of this episode. And there is a white paper on color in digital pathology that’s going to be in the show notes. There is an ebook, why should you choose a medical display for digital pathology? And there’s also more information about Barco MDPC 8127.

And this is a display you guys have for digital pathology. All this information is going to be in the show notes. Thank you so much for joining me today.

Tom ([01:24:47].886)

Thank very much, Alex. enjoyed it. Thanks.

If somebody is interested in contacting you about a monitor, Tom, should they just go to your website or any specific link that we want to share in the show notes?

I’ll provide you with my email address. People can always directly email me. That’s going to be the easiest way, I believe.

Perfect, I’m gonna do that. And thank you so much for sharing your expertise today.

My pleasure.

Monika ([01:25:14].894)

Thank Thank you both. It was nice interacting and had a wonderful discussion. Thank you.