สำเนาบทสนทนา 🔗
WEBVTT
00:00:26.289 --> 00:00:30.129
We are live at Pod Visions and I'm joined by Amy Welch.
00:00:30.289 --> 00:00:30.929
Welcome Amy.
00:00:31.089 --> 00:00:32.129
Thank you so much for having me.
00:00:32.210 --> 00:00:33.009
I'm excited to be here.
00:00:33.329 --> 00:00:35.170
This is the first live stream we've done together.
00:00:35.409 --> 00:00:36.609
Yes, that's true.
00:00:36.769 --> 00:00:41.009
And Amy is responsible for the Nano Zumer scanner line at Hamamatsu.
00:00:41.329 --> 00:00:45.969
So I do the product market managing for Nano Zoomer with Hamamatsu.
00:00:46.209 --> 00:00:47.409
So very exciting.
00:00:47.570 --> 00:00:50.929
We have a lot of products here, new things.
00:00:51.090 --> 00:00:54.609
So what happened is we collected questions.
00:00:54.769 --> 00:00:57.570
This beautiful box was built by me, Dr.
00:00:57.730 --> 00:01:01.170
Alex herself at the FedEx.
00:01:01.890 --> 00:01:02.450
It's an original.
00:01:02.849 --> 00:01:03.489
We have to wrap it up.
00:01:04.450 --> 00:01:05.810
We have to raffle it up.
00:01:06.049 --> 00:01:14.929
But uh as the conference was happening, we were uh gathering beginner questions from the audience uh from participants, right?
00:01:15.010 --> 00:01:19.010
And we have a couple of questions and we're gonna be taking them out and answering them.
00:01:19.090 --> 00:01:30.049
And today uh I'm gonna be the one answering questions is looking at digital slides of the scanner, diagnostic quality as the same.
00:01:30.369 --> 00:01:33.729
Oh, the same diagnostic quality as glass slide.
00:01:33.969 --> 00:01:49.569
So um that is the first question you get from people who are new to digital pathology, from pathologists, from scientists, researchers, but mostly from diagnosing pathologists who are new to digital pathology, they're basically afraid that they're gonna miss something.
00:01:49.810 --> 00:02:12.449
But, and you have, I think, at least three scanners that went through that process, and there is like 10 or more that went through the ethnic clearance uh process, let's call it process, the 510k clearance process, which is basically showing that the slides from the scanner are as good as the slide or non-inferior.
00:02:12.530 --> 00:02:17.650
There's this word of non-inferiority, they're not worse than glass lights.
00:02:17.810 --> 00:02:20.370
Uh, so the answer to this one is yes.
00:02:20.689 --> 00:02:22.610
Um what is changing?
00:02:22.770 --> 00:02:26.930
Not the quality of the images, but uh the way you interact with them.
00:02:27.090 --> 00:02:36.449
So it's not gonna be under the microscope, you're gonna not gonna be sitting and like uh focusing on and off, you're gonna be looking at them on the screen, and that's something you need to get used to.
00:02:36.530 --> 00:02:40.849
But other than that, all the diagnostic information is in the digital slide as well.
00:02:41.169 --> 00:02:41.969
That's a great answer.
00:02:42.129 --> 00:03:01.490
So, to sum it up, there's
Load full transcript
a number of companies here that have shown that equivalency through their FDA submission and clearance to give confidence to pathologists and other users that what they're seeing digitally is the same, just as good as what they would see with class.
00:03:01.889 --> 00:03:02.370
Definitely.
00:03:02.449 --> 00:03:06.049
And uh, which of yours went through the process, can you say?
00:03:06.289 --> 00:03:08.049
So we have three systems now.
00:03:08.129 --> 00:03:19.569
We have our S540 MD, which is our largest, our S360 MD, which was our first FDA 510K clearance, and our S20 MD, which is our small one.
00:03:21.169 --> 00:03:23.810
Um so I want to point something out.
00:03:23.969 --> 00:03:27.009
The numbers actually are the numbers how many slides go in, right?
00:03:27.250 --> 00:03:27.569
Capacity.
00:03:27.889 --> 00:03:29.169
Because it's not always the case.
00:03:29.329 --> 00:03:38.930
I was like, uh, I don't know, there was some other scanner that had 600 something, something, and I'm like, oh, it can scan so many slides, and they're like, no, that's just the number in the name.
00:03:39.090 --> 00:03:43.569
So the nano zoomers actually the number in the name tells you how many slides it scans.
00:03:44.049 --> 00:03:45.250
That's correct.
00:03:45.569 --> 00:03:48.449
Okay, I think uh all right, that's a great first question.
00:03:48.689 --> 00:03:49.409
Yes, right.
00:03:49.569 --> 00:03:51.969
It takes us right into the next.
00:03:52.210 --> 00:03:54.930
Yeah, let's do it right in very very digital pathology process.
00:03:55.329 --> 00:04:00.289
How long does it take to scan a slide and then 15 by 15 area?
00:04:00.530 --> 00:04:05.329
Okay, so I'll start with this, but this one is gonna go to you because it's very scanner specific.
00:04:05.409 --> 00:04:19.889
So 15 by 15 is the little square that uh all scanner manufacturer kind of adopted as something to compare apples to apples because, like, from a lab perspective, pathology perspective, every slide is different.
00:04:20.129 --> 00:04:24.610
So every slide is gonna have a different time of scanning.
00:04:24.769 --> 00:04:34.930
Uh, it depends how much tissue is there, how like easy it is to focus, and like a lot of just these physical properties of the slide determine how fast it's gonna be.
00:04:35.009 --> 00:04:36.530
But then you want to compare, right?
00:04:36.610 --> 00:04:43.490
You don't want to buy a slow one if you need a fast one, or you know, depends on your use case, but this square.
00:04:43.649 --> 00:04:49.329
So, your machines, how fast do they scan this square of 15 by 15 millimeters?
00:04:49.649 --> 00:04:59.810
For our 20, 360, and 540 across the board for the 15 by 15 industry standard takes about 30 seconds per slide.
00:05:00.129 --> 00:05:12.050
But as you said, in real-world practice, variation in tissue size, tissue depth, preparation, that will all change of how long it takes to scan a digital slide.
00:05:12.209 --> 00:05:18.769
No, so I'm just trying to think, okay, what this is one and a half centimeter by one and a half centimeter.
00:05:19.009 --> 00:05:20.370
It's pretty small, yeah.
00:05:20.449 --> 00:05:26.209
It's a small area, but like a biopsy could like that, although they are longer.
00:05:26.769 --> 00:05:28.930
So okay, 30 seconds, yeah.
00:05:29.730 --> 00:05:30.529
But it it varies.
00:05:30.850 --> 00:05:31.009
Yeah.
00:05:31.250 --> 00:05:34.449
Do you think you're you have different sample types you're putting into a scanner?
00:05:34.529 --> 00:05:39.329
Maybe you have some um core needle biopsies and some larger resections.
00:05:39.810 --> 00:05:44.529
So yeah, sections that the one that holds you back.
00:05:44.769 --> 00:05:48.610
What surprised you the most when you started in digital pathology?
00:05:49.009 --> 00:05:49.969
I'm gonna start with you.
00:05:50.129 --> 00:05:51.170
When did you start?
00:05:51.649 --> 00:06:02.290
Um, so it was late 2018, and then when I first got introduced to digital pathology, and it couldn't be more different than what we're seeing in the field right now at that time.
00:06:02.689 --> 00:06:03.810
So, what's worth doing?
00:06:03.889 --> 00:06:05.810
Like describe your situation.
00:06:05.969 --> 00:06:06.610
Where were you?
00:06:06.850 --> 00:06:19.409
So I was working with the scanner, and it was really in kind of a closet back room with a lot of slides piled up to archive, as I think a lot of people were in similar situations at that time.
00:06:19.730 --> 00:06:27.730
And it's been really great to see the transition since that point and the introduction, more work with AI and everything that can be done.
00:06:27.810 --> 00:06:40.290
And so I always think about sitting in that room and scanning a ton of slides to build this repository, and now we see people that are putting their slides in right away.
00:06:40.449 --> 00:06:44.529
That result you get is almost instantaneous, as close as instant as it can be.
00:06:44.769 --> 00:06:49.170
So it's been really great to see that transition in the field as a whole.
00:06:49.409 --> 00:06:51.490
Well, when you started, what was the surprise?
00:06:51.649 --> 00:07:01.009
Like, did you expect it to be faster, slower, not have to sit in the basement, or like when when you learned that you're gonna be doing this?
00:07:01.170 --> 00:07:03.889
What was the the surprise factor?
00:07:05.170 --> 00:07:09.009
I didn't have any expectations going into it.
00:07:09.730 --> 00:07:23.089
I had come from a hospital that wasn't digital, so I thought this is really something new and cool, and you would compare it to now and think, oh, the the scanners were much larger or slower or different.
00:07:23.409 --> 00:07:32.769
At the time, it felt like you were working with some really innovative technology, it was this new era time mode.
00:07:33.250 --> 00:07:37.569
So I had like kind of two-step encounter with digital pathology.
00:07:37.810 --> 00:07:45.969
The first one um was when I
was doing my residency, veterin pathology residence in Berlin, and we had a scanner.
00:07:46.129 --> 00:07:57.730
It was like six-slide scanner, and it was very protected in a specific like corner of the lab, and only two people were allowed to touch it.
00:07:57.969 --> 00:08:09.170
So when there were like some slide conferences, or we were preparing um interesting cases for conferences, then these two people would like collect them and scan them.
00:08:09.250 --> 00:08:18.689
So my first impression of digital pathology was like, oh, this is just for the experience and for the elites, and like we don't even come close to the scanner.
00:08:18.769 --> 00:08:25.490
Like, this is a technology, like some first-year resident from Poland, like don't even think of touching it.
00:08:25.649 --> 00:08:29.889
So that was my first step experience, which like oh, super elite technology.
00:08:30.050 --> 00:08:37.250
And then my first job experience, I actually ended up in a digital pathology company that was doing image analysis.
00:08:37.409 --> 00:08:40.129
And that was there was no glass at all.
00:08:40.290 --> 00:08:41.730
I didn't even have a microscope.
00:08:41.889 --> 00:08:42.210
Wow.
00:08:42.370 --> 00:08:52.129
So that was like totally the opposite extreme, where I came with my pathology knowledge and I was already like immersed in this technology.
00:08:52.290 --> 00:09:05.490
Kind of, and the surprise, I mean, I applied for this job, so I knew it's gonna be digital pathology, so it wasn't the surprise, but the surprise was when I finished this job and went into the real world that it was in-between.
00:09:05.649 --> 00:09:09.490
It was neither super elite nor was it everywhere.
00:09:09.730 --> 00:09:20.049
I was so eager to like use my digital pathology experience in my next job and buy some image analysis projects, and it took some time to get up to speed, and that's you.
00:09:20.450 --> 00:09:21.730
You became the trailblazer.
00:09:22.049 --> 00:09:22.850
Exactly.
00:09:23.409 --> 00:09:25.490
See, actually, you know how I became a trailer.
00:09:25.570 --> 00:09:29.009
I mean, I became a trailblazer by trying to blaze this trail.
00:09:29.250 --> 00:09:36.689
But one time a colleague of mine from the preclinical space, he called me like that, and I'm like, that's what we are.
00:09:36.850 --> 00:09:48.289
That's what we all are in digital pathology because it's no more for the elites, but we still need to blaze the trail to increase the adoption because to leverage all the benefits.
00:09:48.450 --> 00:09:56.370
So that was my I don't know if it's this surprise is surprising enough, but that was my like realization, reality checking.
00:09:56.529 --> 00:10:00.049
Is there something that is for small volume scanning?
00:10:00.210 --> 00:10:03.169
Uh there is one for small volume scanning in this booth.
00:10:03.330 --> 00:10:05.649
Which one is your for small volume scanning?
00:10:05.809 --> 00:10:09.809
So, from our lineup, our smallest scanner is our S20.
00:10:09.970 --> 00:10:17.889
So that holds 20 slides, and there's RUO and MD medical device for both of those, depending on what your use case is.
00:10:18.049 --> 00:10:24.929
But in the industry overall, there are some very small scanners that are one slide, four slides.
00:10:25.090 --> 00:10:28.689
It really depends on what you're looking to do and your goals.
00:10:29.570 --> 00:10:34.929
And MD is not just for medical doctors, it's the medical device that was cleared by the FDA.
00:10:35.009 --> 00:10:40.929
And RUO is researchers only, so uh all the researchers can use that one, uh, the scanners as well.
00:10:41.169 --> 00:10:44.129
Will AI replace pathologists?
00:10:44.610 --> 00:10:58.929
So this is a hallmark question of any digital pathology conference, and probably like after any digital pathology publication that I review online.
00:10:59.490 --> 00:11:03.889
Um, and it's also not specific to our technology.
00:11:04.850 --> 00:11:14.929
Whenever there is a new technology entering, people are afraid, okay, what's gonna happen to how we now like interact with the universe at the like very high level.
00:11:15.090 --> 00:11:18.370
Uh, and in our case, okay, will pathologists be replaced?
00:11:18.450 --> 00:11:20.210
And there is a couple of slogans.
00:11:20.450 --> 00:11:23.970
Um, I just came from a very good talk by Raj Singh.
00:11:24.129 --> 00:11:30.929
Uh he had the keynote or like plenary talk, and there's a couple of slogans, and he had this as a question on his slides.
00:11:31.169 --> 00:11:46.529
There are a couple of slogans like one of the most popular in the pathology space is no AI is not gonna replace pathologists, but it's gonna but um the pathologists who use AI are gonna replace the ones that don't use AI.
00:11:46.689 --> 00:12:14.689
And I on purpose say it's kind of a slogan, and uh I think we're beyond that because this was kind of true at the very beginning of this AI empowerment of pathology, because there you actually had to have some knowledge how to design an image analysis algorithm, how to maybe interact with specific software that does that.
00:12:14.850 --> 00:12:32.450
Now it's all voice-driven, and what he was pointing out that half of this AI or like 70% of this AI you don't even see because this is workflow, it's guiding the machine that's scanning, it's uh, I don't know, gathering all the data that you need to use your brain.
00:12:32.610 --> 00:12:36.769
And his answer to this, and I'm gonna be citing him because he had a very good answer.
00:12:37.009 --> 00:12:45.649
Patholog uh AI is not gonna replace doctors nor pathologists, it's gonna make them more relevant for patients.
00:12:46.210 --> 00:12:46.769
Yes.
00:12:47.009 --> 00:12:59.730
I think I remember hearing in one talk someone saying what it's gonna help do is help a generalist, a general pathologist, it'll help them be as good as an expert in certain areas.
00:12:59.889 --> 00:13:01.169
What are your
thoughts on that?
00:13:01.409 --> 00:13:11.970
So that very much ties into all the AI algorithms that are now available that can run on these slides and um help you diagnose, right?
00:13:12.129 --> 00:13:13.570
So that's one aspect.
00:13:13.730 --> 00:13:20.129
Um the other aspect is that there's gonna be more focused.
00:13:20.289 --> 00:13:24.929
So the other part of this, like, okay, will parts of my job disappear?
00:13:25.169 --> 00:13:33.169
And his answer was yes, it's gonna disappear because you you're no longer will have to count chi-67 positive cells.
00:13:33.250 --> 00:13:36.529
You no longer have to do this whole like out of stack of papers.
00:13:36.689 --> 00:13:40.370
So this will disappear, and obviously I'm super happy that this will disappear.
00:13:40.529 --> 00:13:50.450
Um, but what will what we will put more focus on is gonna be the aspect of translation.
00:13:50.850 --> 00:14:04.610
Translation between um the diagnosis and care pathologist being the translational scientist in the healthcare space, and also uh there's gonna be more interactions with patients.
00:14:04.850 --> 00:14:15.970
The relevance of a doctor to patient as a person and as opposed to to patient as um like their condition is gonna increase.
00:14:16.210 --> 00:14:23.730
Because now, I mean, anybody who has been uh at the doctors, probably anywhere in the world, it's like fast, fast, fast.
00:14:24.049 --> 00:14:53.490
Uh only they have allocated certain amount of minutes to see a patient, and uh I hope that's gonna shift because the time used for any kind of mundane administrative, any kind of like logistics that you need to do is gonna shrink, and that's gonna open time to interact with patients or to do better science, to do more research, like the cognitively valuable aspect of the pathologist's job.
00:14:53.649 --> 00:14:55.809
So that was a convoluted.
00:14:56.689 --> 00:15:00.850
Technology is gonna help us focus more on the human.
00:15:01.169 --> 00:15:05.250
And I think that's what we're missing in general.
00:15:05.409 --> 00:15:16.689
I don't know, that's might be part of a philosophical discussion, but you know, all this like uh looking at the phones, uh interacting with digital instead of in person.
00:15:16.850 --> 00:15:18.929
Like these there are some slogans in coffee shops.
00:15:19.090 --> 00:15:21.330
No, we don't have Wi-Fi, talk to each other.
00:15:21.570 --> 00:15:22.289
I've seen those.
00:15:22.610 --> 00:15:28.289
And there's and here we're in San Diego, so I think they they are leaning a lot into that.
00:15:28.529 --> 00:15:41.330
Um and there's gonna be, and rightly so, because not that it's like something bad, but this over excitement is ending, and now it's gonna be okay, where does it help us, where does it not help us?
00:15:41.490 --> 00:15:44.929
And let's eliminate the stuff that it doesn't help us with.
00:15:45.090 --> 00:16:04.370
Um, and it's not always helps us with human connection, and especially, well, sometimes it does, the the remote calls, remote meetings, remote work, but uh where it doesn't, we now have interacted with it enough to recognize where it doesn't, and then we can just skip that part of it.
00:16:05.009 --> 00:16:08.769
And I think that translates into medicine onto pathology as well.
00:16:08.929 --> 00:16:11.649
What do you think on AI replacing anything?
00:16:11.970 --> 00:16:14.370
I don't think it'll happen with pathologists.
00:16:14.529 --> 00:16:21.490
I think the pathologist is always gonna be there and has a place there, and I think that's also really important for patients.
00:16:21.970 --> 00:16:26.769
They want to know that there's a trusted physician also looking at their samples.
00:16:27.009 --> 00:16:38.769
Yes, technology can help um, you know, maybe with their confidence as well, or or help us do things faster, which with patients, time is so important.
00:16:39.169 --> 00:16:50.769
But I think there is always that place for pathologists, and they're really the experts, and there's certain ways that technology, at least as it is right now, will never be better than their eye for certain things.
00:16:51.009 --> 00:16:53.490
And that's truly a gift of pathologists.
00:16:53.889 --> 00:16:54.929
You know what else?
00:16:55.090 --> 00:17:14.769
Um that's not strictly AI, but something that's gonna bring us closer to patients being part of care, patients feeling better in understanding their diagnosis, is them seeing digital images with the uh with the pathologist consulting on them.
00:17:15.009 --> 00:17:19.090
So there is, I don't know if you've heard of this initiative, they're called pathology clinics.
00:17:19.490 --> 00:17:29.250
And in several hospitals, there are actually uh like clinic hours where pathologists meet with patients.
00:17:29.490 --> 00:17:42.130
The patient can like schedule a visit with a pathologist in person, and they look together at their slide, and the pathologist is explaining the biology and um everything about their diagnosis.
00:17:42.530 --> 00:17:43.170
That's really great.
00:17:43.570 --> 00:17:45.330
I just thought it's the enabler.
00:17:45.810 --> 00:17:59.170
There hasn't been that connection, so much so, between the pathologist and the patient, and bringing those closer together and helping patients understand more of what's going on with themselves and their journey helps empower them.
00:17:59.730 --> 00:18:01.490
And that's really what we're all here to do.
00:18:01.650 --> 00:18:07.650
Except so the very short answer to that question will AI replace pathologists?
00:18:07.890 --> 00:18:10.130
No, it's not gonna replace them.
00:18:10.370 --> 00:18:15.170
How does a full slide scanner turn a glass slide into a digital image?
00:18:15.410 --> 00:18:16.770
Do you want to take that one?
00:18:17.090 --> 00:18:23.250
If you want to start it off high level, and then we can go uh a little bit more scanner-specific.
00:18:23.570 --> 00:18:28.450
But and that goes back to
the first question, which was like, oh, is the image good enough?
00:18:28.610 --> 00:18:32.610
Uh because now it's not under the microscope, now it's coming out of a scanner.
00:18:32.850 --> 00:18:42.530
So um I'm sorry to disappoint you if you were expecting something else, but actually, scanner is a type of microscope inside.
00:18:42.770 --> 00:18:50.130
So the technology that is underlying cross-light imaging and digital pathology is actually light microscopy.
00:18:50.290 --> 00:18:56.370
So that's the same micro technology that I use when I was training in my residency.
00:18:56.610 --> 00:19:08.370
Um and it's good news because um there's this discussion, okay, if people are not working on digital slides, do they need like additional training, courses?
00:19:08.610 --> 00:19:11.810
Uh my take on that is no, it's the same technology.
00:19:11.890 --> 00:19:20.130
It doesn't matter if you were trained on a microscope or on whole slide images, the information is the same, and your expertise is to interpret that information.
00:19:20.290 --> 00:19:21.890
How you acquire this information?
00:19:22.050 --> 00:19:25.970
Well, most of us have worked with a computer and with a digital image.
00:19:26.050 --> 00:19:36.210
Uh, probably like if you are at this conference, you may have seen, might have seen a lot of cell phones uh that I take who is gonna take one right now from our live stream.
00:19:36.370 --> 00:19:41.730
Um so the the hurdle is not big, the technology underlying is the same.
00:19:41.810 --> 00:19:46.290
So there's like a motorized stage objective uh and the slide there.
00:19:46.530 --> 00:19:49.650
Camera camera, exactly, creating this image.
00:19:49.810 --> 00:19:51.890
Anything we want to add to this?
00:19:52.130 --> 00:19:53.490
I think that's exactly right.
00:19:53.650 --> 00:20:02.530
It's uh basically a microscope and housed in a machine to automate the work that a pathologist or someone else would do.
00:20:02.690 --> 00:20:13.490
Moving a slide around a stage, there's an objective, there's a camera, uh, but it's allowing those images to then be digital and be stored for X amount of time or forever.
00:20:13.810 --> 00:20:15.970
So just like to show a demo.
00:20:16.050 --> 00:20:17.730
So that's our glass slide, right?
00:20:17.890 --> 00:20:24.050
Glass slide is gonna be so do the slides move or does the objective move, or can it be both?
00:20:24.690 --> 00:20:25.890
It could be both.
00:20:26.130 --> 00:20:27.330
Usually it's a stage.
00:20:27.730 --> 00:20:28.770
The stage that moves, right?
00:20:29.170 --> 00:20:31.010
The objective stage or the objective moves.
00:20:31.170 --> 00:20:42.690
So you're gonna be moving the slide in a specific pattern, or if there is a different version of that, but the the idea is that you create uh this this image, right?
00:20:42.770 --> 00:20:51.730
And you don't image the whole thing um like at once, and it creates the Google Map kind of zoomable image.
00:20:51.890 --> 00:20:54.770
Uh you image either tiles or lines.
00:20:54.930 --> 00:20:56.290
You guys do tiles, right?
00:20:56.450 --> 00:20:57.330
Yes, that's correct.
00:20:57.730 --> 00:21:13.330
So you tiles little squares, and then inside the scanner, there is a software that stitches them together and makes this zoomable experience possible so that you can interact with the digital slide as you would with a glass slide under the microscope.
00:21:13.810 --> 00:21:16.130
So no worries, digital pathology.
00:21:16.290 --> 00:21:19.090
If you're trained on the microscope, you're good.
00:21:19.170 --> 00:21:21.970
You're good as like you're already trained for digital pathology.
00:21:22.930 --> 00:21:24.210
These are great questions.
00:21:24.770 --> 00:21:26.770
Our guys are asking them.
00:21:27.090 --> 00:21:29.410
I'm glad they were so animated.
00:21:29.490 --> 00:21:36.850
I was motivating them very intensely at the booth to put questions in, I have to say that.
00:21:37.090 --> 00:21:41.730
So here I think is for our geared towards basic getting started.
00:21:41.890 --> 00:21:43.410
What is digital pathology?
00:21:43.970 --> 00:21:48.290
So that's even higher level than what we just said, how the digital slide is created.
00:21:48.370 --> 00:21:50.610
But there is a little bit more components to that.
00:21:50.770 --> 00:22:01.170
So digital pathology is the technology that enables the creation of digital images from glass slides.
00:22:02.130 --> 00:22:13.970
No, actually, let's take it a step back and be broader from tissue from tissue because I think in the future it may not be glass slide anymore.
00:22:14.210 --> 00:22:24.770
But basically the technology that enables seeing tissue, physical tissue on a cellular level in a digital format.
00:22:25.090 --> 00:22:58.450
So so the most Classical way, the way that was cleared by the FBA is what we're talking about here right now, whole slide imaging, creating the digital images from uh glass slides, and then there is like let's say this creation of the digital image is the central point, and then you have a lot of like rays going out because it's gone because now we have AI, and now we also have um so AI is gonna be downstream.
00:22:58.530 --> 00:23:13.250
You have the digital uh image, and before the glass slide was a data point, there was a diagnosis, you the pathologist would diagnose it or like the description of tissue, you had a report, and that was your data point.
00:23:13.490 --> 00:23:16.850
Now, because it's digital, it's a data source.
00:23:17.250 --> 00:23:23.250
So because it's a data source, you can extract a lot more than just this one information.
00:23:23.410 --> 00:23:53.090
There's a lot of technologies, AI being one of them, but like AI is obviously an umbrella word, and but all these cutting-edge new technologies are able to extract more and more information from that digital image, be it like quantification, maybe some molecular predictions, and and different things, like
all the vendors are actually um addressing a different point of what you can do with this data.
00:23:53.250 --> 00:23:56.370
You guys sit in the in this central spot.
00:23:56.530 --> 00:24:19.170
But then now, because we realized, well, we didn't realize that it's like wide no garbage in, garbage out, and to not have garbage in, you have to go into the pre-analytical and optimize so that the digital slides and the AI are better and more like consistent, more standardized.
00:24:19.570 --> 00:24:32.770
Uh, and there is a lot of new companies focusing on that, so they kind of are going from the digital pathology uh central spot of digitizing uh into the other direction.
00:24:32.850 --> 00:24:42.770
How do we make the samples kind of I don't know, straighter on the slide or uh provide more tissue on the slide?
00:24:42.930 --> 00:24:47.250
So yeah, long answer to what is digital pathology.
00:24:47.490 --> 00:24:53.410
I really like something you said, thinking now of the image as the data source rather than data point.
00:24:53.890 --> 00:24:58.610
So basically, that's what it is, and then what can you do with this data source?
00:24:58.770 --> 00:25:08.610
Whatever technology and imagination allows you to do, uh, whatever your purpose is, the topic of the conference is from innovation to purpose.
00:25:08.930 --> 00:25:09.090
Yes.
00:25:09.330 --> 00:25:22.690
So whatever innovative ideas you have and the technology to do it, and whatever the purpose of this technology is gonna be for you, this digital version of information from tissue is gonna be the enabler.
00:25:22.850 --> 00:25:23.730
So I'm a fan.
00:25:24.050 --> 00:25:24.850
Of course, war.
00:25:25.890 --> 00:25:27.090
I think that's a great answer.
00:25:27.490 --> 00:25:28.370
Anything you want to add?
00:25:29.410 --> 00:25:32.050
I think you hit about everything we could hit.
00:25:32.290 --> 00:25:32.930
Yeah.
00:25:33.330 --> 00:25:33.650
Okay.
00:25:34.290 --> 00:25:36.930
I think it's just an it's an exciting feel.
00:25:37.570 --> 00:25:42.930
It is, and we have a question what will be the future of digital pathology?
00:25:43.090 --> 00:25:46.210
When will it be in 25 years?
00:25:46.450 --> 00:25:46.850
Wow.
00:25:47.090 --> 00:25:59.490
25 years is hard to think that far in the future because of how we've advanced so much in technology, even just in five years or every year, and we're seeing that happen so much faster now with AI.
00:25:59.730 --> 00:26:03.490
Um, I'm gonna tell you something.
00:26:03.730 --> 00:26:05.410
Um that's yeah.
00:26:06.690 --> 00:26:14.130
It's like like you said, you cannot really predict because it it um changes fast, but then other stuff changes so slow.
00:26:14.290 --> 00:26:28.210
Like five years ago, uh people would say, Oh, everybody's gonna be digitized in five years, everybody will have a scanner, and this did this doesn't happen, but then nobody predicted that oh, you can now have uh molecular terror tests based on HED.
00:26:28.370
--> 00:26:29.730
So, you know, different aspects.
00:26:29.890 --> 00:26:38.450
But when I was talking to your team, I was um surprised where hello, we are doing a live stream.
00:26:38.850 --> 00:26:41.330
Oh god, you can you can you want to be introduced?
00:26:42.210 --> 00:26:43.250
No, are you sure?
00:26:43.330 --> 00:26:51.090
You can you can join and just say hi to the digital pathology trailblazers of the digital pathology podcast.
00:26:51.170 --> 00:26:54.050
So go ahead, your name and yeah, introduce yourself.
00:26:54.450 --> 00:27:13.090
My name is Elizabeth Lakely, and I am here because my husband is the president, and it's really exciting to see the ways that digital pathology will be able to improve and speed up patient care because we need more pathologists, and using the AI tools that are being shown is a really excellent way to do it.
00:27:13.170 --> 00:27:15.570
So it's exciting to see all the changes.
00:27:15.650 --> 00:27:23.170
So your husband was actually actually about to join us, but because he's the president of the DPA, he was busy.
00:27:24.130 --> 00:27:26.050
So you got stuck with me instead.
00:27:26.370 --> 00:27:36.290
No, we are honored exactly orchestrated it together with Amy, but we're honored to uh we are honored to have you as his representative by accident.
00:27:36.610 --> 00:27:37.810
This is a great cameo.
00:27:38.930 --> 00:27:52.930
But no, it's been really exciting, and part of our part of his job this year has been to go to um across the puddle to Graz and Austria for um ECP and um to Stockholm.
00:27:53.010 --> 00:27:54.050
And so it's really exciting.
00:27:54.210 --> 00:27:58.530
See, these aren't just efforts taking place in the United States, it's really worldwide.
00:27:59.090 --> 00:28:04.690
And so when all of us working together, we'll be able to accomplish some really amazing things for digital pathology.
00:28:05.330 --> 00:28:05.970
Got the questions.
00:28:06.130 --> 00:28:09.810
What do you think digital pathology would be in 25 years in the future?
00:28:10.050 --> 00:28:15.490
What do I think what digital pathology is gonna look like, or pathology is gonna look like in the future?
00:28:15.810 --> 00:28:38.610
I think you're gonna see a lot of AI, particularly serving under cert areas, because there's a critical need for pathologists, and I think a lot of people here's Amy's microphone haven't realized how interesting pathology could be because of AI and the computational tools and uh tools like the Nano Zoom where to scan a slide so you don't have to actually send a slide through you know the post.
00:28:39.010 --> 00:28:55.410
So I think you'll see a lot of um you know AI computer assisted work, but you're still gonna need human pathologists because there are gonna be slides that don't make sense to a computer because they haven't because AI hasn't seen enough of those cases.
00:28:55.730 --> 00:29:07.650
So you're gonna you're gonna need you know definitely a layer of human beings looking at slides, but I think the in 25 years it's gonna really support having services to the outlying areas.
00:29:07.890 --> 00:29:16.210
Like the speaker who is a keynote from Kenya was talking about how they have, I think, one to three pathologists for every million people.
00:29:16.450 --> 00:29:16.770
Wow.
00:29:17.090 --> 00:29:22.290
Which is a I thought we had a few in Poland for 35 million.
00:29:22.450 --> 00:29:24.130
We had like maybe five tons.
00:29:24.930 --> 00:29:25.810
We're lucky.
00:29:26.050 --> 00:29:26.370
Yeah.
00:29:26.530 --> 00:29:28.690
And in the States it's a much lower ratio.
00:29:28.850 --> 00:29:35.650
So you have areas that don't have access, so they can't find out, you know, what I have, is this something that can be cured?
00:29:35.810 --> 00:29:36.930
What stage am I?
00:29:37.170 --> 00:29:44.130
And so hopefully using the networks and using these tools, we're gonna really speed and improve patient care.
00:29:45.090 --> 00:29:50.050
And I think add some really interesting facets to pathology which haven't been there before.
00:29:50.930 --> 00:29:56.610
It's not just slides and microscopes, it's slides, microscopes, computers, and you know, digital imaging.
00:29:56.690 --> 00:29:59.330
It's a whole different ballgame than it was 10 years ago.
00:29:59.650 --> 00:30:00.930
Very much, very much.
00:30:01.170 --> 00:30:12.530
One thing that I think is gonna happen, and that's maybe my wishful thinking, but I am waiting for the direct to digital imaging of tissue at scale.
00:30:13.970 --> 00:30:25.650
I want us not to be digitizing the analog anymore, but imaging tissue directly, like direct to digital, like radiology does.
00:30:25.970 --> 00:30:27.330
Or like the dentistry does.
00:30:27.410 --> 00:30:31.730
I mean, when was the last time you went to a dentist and they were taking old-fashioned x-ray?
00:30:32.370 --> 00:30:32.930
Digital image.
00:30:33.170 --> 00:30:34.930
Yes, digital image.
00:30:35.090 --> 00:30:42.290
Uh so obviously it's um it's gonna be advancement in optics and photonics, probably.
00:30:42.450 --> 00:30:44.290
And that's I'm emphasizing photonics.
00:30:44.530 --> 00:30:45.890
And you know that as well.
00:30:46.210 --> 00:30:50.770
Yes, because uh at Hamamatsu, Photon is their business, it's right.
00:30:50.930 --> 00:30:53.570
That's like our mini inside joke.
00:30:53.970 --> 00:30:56.290
And Elizabeth actually worked at Hamamatsu.
00:30:56.530 --> 00:30:56.930
You did?
00:30:57.090 --> 00:30:57.330
Yes.
00:30:57.570 --> 00:30:58.930
Oh my goodness, so wow.
00:30:59.330 --> 00:31:03.090
I was part of the first team that built a joint EUS website.
00:31:03.730 --> 00:31:08.530
And um, it was really it's it's exciting to work with all of the different countries.
00:31:08.770 --> 00:31:10.610
And then um, oops, sorry.
00:31:10.690 --> 00:31:19.570
Um so I worked on a joint EUS project for the website, and then ultimately I started my own firm working with um
manufacturing and doing strategic planning.
00:31:19.650 --> 00:31:25.170
So it was uh a tremendous learning opportunity that you know made my life pretty interesting and fun.
00:31:25.410 --> 00:31:28.210
Amazing and met some wonderful people, right?
00:31:29.330 --> 00:31:34.530
I mean, do you have like something that you would like to see in digital pathology in 25 years?
00:31:34.690 --> 00:31:38.690
I love the the bringing it to underserved areas.
00:31:39.010 --> 00:31:56.370
I think that kind of what we've been talking about in the theme here today is really bringing it back more to the patients and the humans and using that technology to advance everything for all of us who are getting the end result from pathology and digital pathology.
00:31:56.770 --> 00:32:04.290
I I think it's hard to conceptualize, as we were saying, where it's gonna go in 25 years, but I'm excited to see where it goes.
00:32:04.690 --> 00:32:11.410
I think we're like uh inclined to overestimate what we know, and then we don't know what we don't know.
00:32:11.570 --> 00:32:19.890
So sometimes the things that we don't know go faster because there's less friction than the things that we're like expecting to happen fast.
00:32:20.050 --> 00:32:24.370
So uh I don't think we're giving very short answers to these questions.
00:32:24.930 --> 00:32:31.810
Well, I was just thinking that 25 years ago you had to have an internet address to go anywhere on the web, really.
00:32:31.970 --> 00:32:37.250
I mean, the internet was very much in its infancy, and now look at all the things you can do.
00:32:37.410 --> 00:32:38.450
It's really extraordinary.
00:32:38.610 --> 00:32:41.490
I know from your phone, like phone wasn't a thing.
00:32:41.730 --> 00:32:42.210
Right, yeah.
00:32:42.530 --> 00:32:43.970
I mean, phone was not the model.
00:32:44.290 --> 00:32:47.490
We all have a supercomputer in your palm or your back pocket.
00:32:48.290 --> 00:32:53.090
Okay, this one is what are the parts of digital pathology workflow?
00:32:53.490 --> 00:32:56.530
So I hope I will not miss any.
00:32:57.090 --> 00:33:00.930
And uh let's start with our hub with the scanning.
00:33:01.170 --> 00:33:01.490
Yeah.
00:33:02.050 --> 00:33:06.450
That's gonna be in the center of digital pathology workflow.
00:33:06.610 --> 00:33:08.450
And um RB specs.
00:33:09.010 --> 00:33:15.010
I I posted it on LinkedIn that oh, everybody says it's not just a scanner, but actually it starts with a scanner.
00:33:15.250 --> 00:33:17.010
So that's why this is where we're gonna start.
00:33:17.090 --> 00:33:20.050
So scanning, digitizing images.
00:33:21.490 --> 00:33:35.010
Um before, let's do before everything that is part of the normal pathology workflow currently, because we are still working uh with glass slides, is then part of digital pathology workflow.
00:33:35.090 --> 00:33:42.050
So everything that was in the lab before, the the crossing, cutting tissue, staining, making these slides.
00:33:42.130 --> 00:33:47.090
Then we have our digitization platform, and then everything that comes after.
00:33:47.330 --> 00:33:50.370
But let's focus on looking at these slides.
00:33:50.530 --> 00:34:06.290
So it's gonna be some kind of viewing software, image management software, so that it's not just so that it's connected to other systems, and then it's gonna be um tools to analyze the images.
00:34:06.450 --> 00:34:18.130
So everything, so we have digitizing the images, then viewing the images, um, managing the images, and then doing something other than just looking at them with the images.
00:34:18.769 --> 00:34:21.409
And that's gonna be where the AIs live.
00:34:21.730 --> 00:34:33.489
And well, not on because the AIs are gonna live before as well to propagate the digital data through the whole pipeline, through the whole digital pathology workflow.
00:34:33.730 --> 00:34:36.289
So, but but how I like, let's do high-level.
00:34:36.450 --> 00:34:58.930
We have the pre-analytics, we have the digitization, we have viewing of the images, management of the images, and analyzing the images in the broadest way because it can be like basic image analysis, it can be advanced AI and predictions uh or something else that we don't even know what that is.
00:34:59.090 --> 00:35:05.329
And on top of that, now we have voice interaction with images, but this all relies on analyzing them in some way.
00:35:05.490 --> 00:35:07.970
So I hope that was concise enough.
00:35:08.130 --> 00:35:09.090
This stuff makes sense.
00:35:10.289 --> 00:35:10.849
Josh, please.
00:35:11.730 --> 00:35:12.450
I'm just gonna sign off.
00:35:12.530 --> 00:35:13.010
Thanks so much.
00:35:13.650 --> 00:35:16.130
Elizabeth, you're amazing.
00:35:16.289 --> 00:35:22.530
This is like the beauty of live streams, and you were brave enough to stay on and actually help assess your questions.
00:35:22.690 --> 00:35:23.170
I appreciate that.
00:35:24.210 --> 00:35:26.849
Thanks so much, and uh hope to see you all around again sometime.
00:35:27.010 --> 00:35:28.289
Yes, okay, thank you.
00:35:29.410 --> 00:35:29.730
Bye.
00:35:29.970 --> 00:35:37.890
Okay, we have one more question, and then what should someone new to digital pathology learn first?
00:35:38.610 --> 00:35:44.130
I think somebody new to digital pathology first should learn the language.
00:35:44.369 --> 00:35:45.250
What does that mean?
00:35:45.410 --> 00:35:48.450
The language of digital pathology and why?
00:35:48.769 --> 00:36:04.369
Because digital pathology is a multidisciplinary field, and to work in any multidisciplinary field, you need to be able to talk, inter, like communicate precisely with people who have expertise different than yours.
00:36:04.530 --> 00:36:14.050
So I'm gonna give you my perspective on that from a pathology perspective, and you can give like your perspective on when new people are joining, maybe from like more technical side.
00:36:14.289 --> 00:36:24.690
I joined the digital pathology world as the
pathologist, and uh before I got humbled, I was like, oh my goodness, I'm the expert.
00:36:24.769 --> 00:36:29.570
I can like show all the cells and recognize where the macrophages and this and that.
00:36:29.809 --> 00:36:35.890
And then my um image analysis colleagues would would ask me, okay, like then what do you recommend?
00:36:35.970 --> 00:36:38.690
How should we design this uh this image analysis?
00:36:38.769 --> 00:36:44.289
Like, should we use this type of segmentation or should we just count to do object detection?
00:36:44.369 --> 00:36:55.809
And I'm like, I don't know, like make these cells the shape, make this analysis, the shape of the cell, and make it as precise and as accurate as possible.
00:36:55.970 --> 00:36:59.250
And they were like, but we need to just count them, we don't need the shape.
00:36:59.329 --> 00:37:01.650
And I'm like, that's true, you don't need the shape.
00:37:01.809 --> 00:37:18.769
So my uh part of the language I needed to learn in that particular situation, and then you know it was uh expanded to different areas of digital pathology, was understanding how computer vision principles translate into what the pathology problems are.
00:37:19.010 --> 00:37:24.690
That uh if you're counting chi-67, you actually don't need instance segmentation for this.
00:37:24.849 --> 00:37:31.410
You can just do object detection and you're gonna solve the problem uh with an a lot easier computational method.
00:37:31.570 --> 00:37:37.410
So that was the part of the language that I needed to learn to be able to contribute with my expertise.
00:37:37.650 --> 00:37:44.930
And then my image analysis colleagues, they needed to learn a little bit about the tissue so that they could tell me, okay, hey, is this real?
00:37:45.090 --> 00:37:46.210
Is this necrosis?
00:37:46.369 --> 00:37:48.130
Is this uh inflammatory cells?
00:37:48.289 --> 00:37:52.130
Where does the tumor end uh what is the tumor margin?
00:37:52.210 --> 00:37:55.250
So and they had a little bit of a learning curve, but me too.
00:37:55.490 --> 00:38:02.289
I was I was an expert in pathology, but there's a lot more than just the pathology aspect in digital pathology.
00:38:03.809 --> 00:38:08.930
And uh so my experience, I had first started in a gross room in a pathology lab.
00:38:09.170 --> 00:38:14.690
Okay, so I knew a lot about the tissue and what was coming out, what was even upstream of the slide.
00:38:15.010 --> 00:38:23.090
But then coming to hardware and getting immersed in digital pathology, there's a lot to learn about the scanner itself and all the components and all of that.
00:38:23.329 --> 00:38:36.530
And then, like you're saying, and with a pathologist who's having that end result, either looking at uh glass, microscope, or digital image, that's almost a whole other world and a whole other lingo to learn.
00:38:36.769 --> 00:38:47.890
Um in your perspective, in your experience, what is something that maybe is commonly said in the industry that would make a pathologist think, hmm, we probably wouldn't say it that way.
00:38:48.530 --> 00:38:49.410
Oh my goodness.
00:38:50.050 --> 00:38:50.930
I don't know.
00:38:51.570 --> 00:38:53.090
I don't know, let me think.
00:38:53.570 --> 00:38:55.250
It's gonna come to me.
00:38:55.490 --> 00:39:08.369
Um maybe I'm gonna tell like a different story uh where I've seen um people in the industry put this in practice in the correct way.
00:39:08.530 --> 00:39:27.650
So I have had I've seen members of uh the the scanning scanner team of your team talk to pathologists in one way and then talk to people from engineering background in a totally different way, and they were very fluent in presenting this uh to people.
00:39:27.809 --> 00:39:28.849
So oh, I know.
00:39:29.010 --> 00:39:34.450
So for example, um when a pathologist asks, okay, how fast is the scanner?
00:39:34.769 --> 00:39:46.690
I would like if somebody would tell me, oh, 15 by 15 is uh 30 seconds, if I was not familiar that this is the reference thing, I'm like, what is what does that mean to you?
00:39:46.849 --> 00:39:57.490
Yeah, I would say I would um for example, a more informative would be, oh, like a full-slight resection will take so much, and if uh prostate biopsy is gonna take so much.
00:39:57.890 --> 00:39:59.250
Or like a range, right?
00:39:59.410 --> 00:40:11.329
But if somebody would throw at me 15 by 15, it would be like I have no idea which of my samples are 15 by 15, and they're not square, so why are you even saying 15 by 15?
00:40:11.570 --> 00:40:12.369
I think we're done.
00:40:12.610 --> 00:40:13.090
I think that's it.
00:40:13.170 --> 00:40:15.010
These have been really great questions.
00:40:15.329 --> 00:40:24.050
If there's anyone that's new to pathology that's looking for more information, where can they go to find more about the basics on digital paths?
00:40:24.769 --> 00:40:25.170
Very much.
00:40:25.250 --> 00:40:31.010
See, I don't have my book, otherwise, I would uh show the book because I gave it to one of the of your team members.
00:40:31.490 --> 00:40:36.289
But um, I am working on the second edition of Digital Pathology One on One.
00:40:36.450 --> 00:40:40.050
All you need to know to start or continue your digital pathology journey.
00:40:40.210 --> 00:40:44.610
You will be able to find it on my website, digitalpathologyplace.com.
00:40:45.170 --> 00:40:53.970
And I am keeping my fingers crossed for myself to finish the second edition because there's a lot we need to add to the first one.
00:40:54.130 --> 00:41:01.170
All the AI revolution happened between the last edition from 2023 and now 2026.
00:41:01.250 --> 00:41:02.050
And what this happened.
00:41:02.450 --> 00:41:04.450
Very exciting, and congratulations.
00:41:04.690 --> 00:41:05.490
Thank you so much.
00:41:05.650 --> 00:41:07.490
I can't wait to see the next edition.
00:41:07.809 --> 00:41:10.369
Maybe at next conference, or I'm gonna bring some.
00:41:10.530 --> 00:41:14.930
Thank you so much for co-hosting this with me and for asking me the question.
00:41:15.730 --> 00:41:20.530
It's been great, and thank you all for joining in and having as a guest.
00:41:20.849 --> 00:41:23.170
The questions, unfortunately, we need to wrap this up.
00:41:23.329 --> 00:41:30.690
We have questions in the comments, and I'm gonna be looking with my team at the questions in the comments, and we're gonna answer you even though it wasn't live.
00:41:30.930 --> 00:41:33.970
Thank you so much for joining, and we'll talk to you in the next episode.
00:41:34.130 --> 00:41:34.530
Bye.
00:41:34.690 --> 00:41:35.410
Thanks.