184: Digital Pathology Guidelines: What Every Lab Must Get Right

What actually needs to be in place before digital pathology can replace the microscope?

In this episode of DigiPath Digest, I walk through the 2026 Polish Society of Pathologists guidelines and translate them into practical steps for real pathology labs. This isn’t theory. It’s about hardware fidelity, data integrity, validation, and AI integration — and what each of these actually requires in daily workflow.

We talk about scanner resolution standards (≤0.26 μm per pixel), 4K monitor calibration, visually lossless compression (20:1), scalable storage, pathologist-driven validation, and what “non-inferiority” truly means.

Digital pathology is not just a change of medium. It’s an operational shift.

Episode Highlights

[00:02] Community & growth
1,600+ new newsletter subscribers, 10,000+ Facebook members, and free Digital Pathology 101 book access.

[07:20] The 4 pillars of adoption
Hardware fidelity · Data integrity · Clinical validation · Future integration.

[08:30] Hardware requirements
40x equivalent scanning (≤0.26 μm/px), 4K monitors, >300 cd/m² luminance, 10-bit color depth.

[12:00] Workflow & throughput
200–300 slides/day per scanner, automated focus control, urgent case prioritization.

[17:25] Storage & archiving
~1 GB per slide. Active archive (6–24 months). Long-term retention (10–20 years). GDPR compliance & TLS encryption.

[23:09] Validation philosophy
Pathologist-centered validation.
Two phases:
• Familiarization (~20 retrospective cases)
• Dual review with discrepancy tracking
Goal: digital must be non-inferior to glass.

[29:03] AI in digital pathology
AI supports quantification (Ki-67, HER2, ER/PR, PD-L1), tumor detection, and future multimodal predictions — but pathologists remain central.

[33:26] Intraoperative telepathology
<5-minute scan-to-view time.
Minimum 100 Mbps upload.
Redundancy and safety protocols required.

[34:50] Can digital cameras replace scanners?
Hybrid workflows exist. Regulatory compliance still applies.

[38:19] Adoption checklist summary
Certified scanners (CE-IVD/FDA), calibrated monitors, scalable storage, phased validation, and documented QC.

Key Takeaways

• Digital pathology adoption is a structured process — not just buying a scanner.
• Validation is individualized and tissue-specific.
• Infrastructure and quality control are as important as image quality.
• AI enhances reproducibility and quantification but does not replace pathologists.
• Regulatory compliance and data governance are non-negotiable.

Transcript

00:00:02
Aleks: Welcome my trailblazers. Happy pre Valentine's Day. Welcome to today's Digipath [music] Digest. Because it's pre Valentine's, I have music playing. Uh experimenting with features a little bit and waiting for you to show up. Okay, I'm seeing you welcoming you in the chat as [music] well. Let me know in the chat. Well, my spelling [music] is not fantastic. Let me know in the chat where you're tuning in from. H it's 6:01 in [music] Pennsylvania, 13th of February, day, Monday before Valentine's. Let me know

00:00:44
um that you hear me well. Let me know where you're [music] tuning in from. And I'm going to start with a few updates because this week was [music] uh pretty special. You guys showed me so much love. And by so much love, I mean we have like over 1,600 new trailblazers [music] joining the newsletter because um I decided that even though I'm still working on the book uh updates and I'm going to give you [music] the code to the book. So if you don't have the book, you can get the book right now. Now that

00:01:21
the music stopped playing, I can also take off the headphones. Um, yeah, you should see the code to the book. Uh, so if you don't have this book yet in the digital form because the code is for the digital form, click on the code. Uh, no, you don't click. You scan it. You scan it. It's a QR code and you can get the book. So, a crazy thing happened. Crazy thing happened. Okay. And you guys are hearing me well. This is fantastic. and greetings. Greetings from P from Fairfield, Pennsylvania to

00:01:55
Talin. Okay, so 1,600 people joined the newsletter, the digital pathology newsletter. This is amazing because I decided that hey this book even though I'm working on the updates and you know because if it's not your first time here if it's your first time I'm still working on the updates but um the like the stuff that's in the book is still super valuable for people starting their digital pathology journey. So I said okay let me just put it on social media again and 1,600 people wanted the book.

00:02:30
Uh I that means I didn't do a good job giving it to you in the first place. H and that means that we have now like 10,000 over 10,000 trailblazers on Facebook. Uh and yeah, the the newsletter grew substantially. So if you're just joining, let me know in the chat. Um QR codes the book for you. Where is it? here the book if you don't have the book yet. So that's one thing and the big part for you. Thank you so much because uh you make this uh technology move forward. Basically um I'm teaching it to

00:03:15
people who will take it and actually implement it whenever they get the chance. Um and even if you don't yet have a scanner, if you're still working with a microscope, it all starts with awareness and knowledge. So this can be your first step digital pathology 101 book. Um and then we decided and by we I mean those who joined the live stream last week that we're going to talk about the guidelines published by Polish Society of Pathologists. Um, I'm going to put the QR code for the

00:03:57
book later as well if you want to check it out if you don't have it yet. And then I have a little surprise or like a little special Valentine offer for you as well. You might have seen it on social media h but maybe not. So I'm going to show it to you here as well. But today is the we're talking about the guidelines for the adoption of digital pathology in clinical pathology unit recommended by the Polish Society of Pathologists. And if you don't practice in Poland, it doesn't really matter. I

00:04:27
obviously am super excited about this being put out in diagnostic pathology by a Polish group. And actually last week I h had the honor and the chance to talk to one of the authors uh Vto Dresnner and out of this conversation because obviously I want to feature what's going on in Poland um there is some initiative that I'm going to be embarking on h that is going to include the authors from uh this particular paper the first author wukau wukau Schielberg as well but what I want to do is to create a

00:05:06
podcast series called digital pathology. I'm of course not showing myself. I want to create this new podcast series where I show the trailblazers from each country. And I probably will will start with my country of origin and then I'm going to invite some people from my country of uh residence right now the US. But uh what I'm going to do I'm going to prepare a kind of structured format because I want to show what's happening in the world right uh we get publications not from all over the world

00:05:51
but I do want to show um what's around the world. So let me know what you think about this idea. Do you want pathology uh around digital pathology around the world or you say me not really I'd rather do more papers. You let me know in the chat what you would like more of and we're going to go back to our publication. H and actually I prepared a presentation for you. I used Notebook LM. Notebook LM is my new best friend from AI tools, especially for um paper summarization, paper uh querying and all

00:06:36
the these different things. It's optimized for academic work. So, I use it a lot and I thought I'm going to use it for this presentation as well. If you're just joining, welcome. And let me know if you want digital pathology around the world as a podcast series. And if you know some people, send them this episode that where I'm talking about this and soon enough there's going to be a framework and like an intake form. But without further ado, let's go into operational shoot sorry

00:07:20
why does it not draw nicely okay doesn't matter operationalizing digital pathology a practical implementation guide technical standards validation protocols and workflow integration based on the 2026 Polish society of pathologist guidelines which is fantastic because we are now just at the beginning of 2026 and we have these guidelines um Schillberg at all in diagnostic pathology presentation created with the help of notebook LM I highly recommend um also if you would like a tutorial how to make this

00:07:55
presentation with um notebook LM which is going to be like maybe five minute tutorial let me know in the comments that you'd like to the tutorial and um in these guidelines we have the four pillars of adoption because transitioning from Halllight imaging requires satisfying strict technical and procedural criter criteria to ensure diagnostic non-inferiority. And we're going to talk about this non- inferiority uh in kind of like philosophical validation terms. But these four pillars are hardware

00:08:30
fidelity, data integrity, clinical validation, and future integration. We're going to dive into each one of them, but I can show it to you better. So, um, scanners must match the 40x optical objectives. 40x for anatomic pathologists. This is what we're using. And uh this has to be great uh equal or greater of 0 26 micrometer per pixel which with most of the scanners it is 0.25. So uh whichever scanner you choose you need to check that but u you should be covered. And then also uh we uh monitors must must be

00:09:21
hardware calibrated for color consistency. We talked on the podcast about color consistency. Um what about data integrity? Well, uh the infrastructure must handle high throughput around 1 gigabyte per slide. It can vary, excuse me, because um they can be bigger or they can be smaller, right? with a visually lossless comp visually lossless compression um 20 to1 and secure archiving clinical validation is a two-phase workflow. We're going to dive into it. H that is divided into familiarization and

00:10:04
dual run and to validate the individual pathologist and not just the machine and uh the leveraging the digital medium for AI assisted quantification. Basically, this is like the superpower of digital pathology. And uh we're going to see tests uh that are based on digital quantification of markers that actually cannot be quantified uh by a pathologist visually. It can be confirmed but not really quantified. And I'm talking about the tropto that um Astroenica was working on and is still working on

00:10:42
and is currently available for research use. Um but this like this future integration this is something I am super passionate about because you cannot do it without digital pathology. You can diagnose without digital pathology. You just do it on the microscope with glass slides. All good. but to incorporate precision medicine um soon enough this digital is not going to be an option. So let's talk about the scanner resolution over uh sorry like less than it has to be more detailed than uh 0.25 25 micron per

00:11:22
pixel less or equal to this is critical for diagnostic uh fidelity and we would like at least 200 two to 300 slides per day per device. This is going to depend on the lab because uh you may need more you can need less maybe if you're only supplementing your workflow with digital then uh you can have a less high throughput scanner. Um but if you're using it for like your general diagnostic work, this is what they suggest. And focus control should be automated with dynamic refocusing or ZStack automated for workflow optimal

00:12:00
optimal workflow because um if it can automatically focus then you don't have to do it manually. But in case it doesn't focus well you need to have the option to do it manually as well. and mapping full slide capture excluding label to prevent missed tissue. So it goes well it doesn't go the the the slide goes into the scanner usually uh and uh there is tissue recognition so everything is scanned. I love this. This is notebook LM version of a scanner but actually it doesn't look so far from

00:12:36
reality. Yeah. Anyway, uh but it looks the the scanner basically is a microscope in a box that scans slides automatically. And oh, we have people from Maryland. Silver Spring. Fantastic. Thank you for joining. It's early for you. I know. I appreciate all those who uh wake up early or stay up late. Um so yeah, let's just smile at the scanner. Everybody chooses their own scanner. But basically when you choose your scanner, you look okay this res at at these parameters at the resolution uh throughput focus

00:13:16
control mapping and um for their workflow requirement is priority queuing features um should be mandatory for handling urgent and intraoperative cases within batch workflows. So it means that you need to have an option to put in a priority case. It can be um so that's a workflow requirement right it's not a scanner requirement if you use a different scanner somehow integrated you can do that but uh it needs to be possible now output specifications the diagnostic station see if I capture

00:13:59
everything I wish it would go like full screen okay that's the closest to full screen. The diagnostic station, the diagnostic station is your monitor and then your computer, right? And you're going to be staring at it for a long time. So, it has to be good for diagnostics and good for you as well so that you can do good diagnostics or your pathologist can do good diagnostics. Um, so the resolution 4K ultra HD preferred. Uh, and I think it can be a little bit less than that. Uh but that's what they say in the

00:14:38
guidelines. 4K is good. Luminescence over 300 candelas per uh meter squared. Uh contrast,000 to1 hardwarebased calibration 10 bit color depth. So they require calibration of monitors. And again, I will send you back to this podcast episode that we had with Tom Kimpe and um Monica Lambasini. They talked extensively about color from the engineering standpoint and from the diagnostic pathologist standpoint and they go deep into why color is important and why actually you should calibrate. [snorts]

00:15:22
So we also have specifications for the computers which is fantastic because I was recently um submitting a publication about our validation and by our I mean Charles River Laboratory's validation of our system for primary and what they were and I was focusing on this pathology component where like how did about our philosophy about the non-inferiority and the reviewers came back like why are you not talking about specs And sure enough, I had to put uh the specs back in. So, if we look at this uh we have

00:15:58
specifications for CPU, i7, Ryzen 7 or better, GPU dedicated over 8 GB V RAM essential for smooth panning. So, um let me do some panning and see how smooth is our PDF. Can I do that? No, I cannot do that. Let me check. No, I wanted to do a dramatic demonstration of smooth panning [gasps and laughter] but um you can look at my videos. I recently I published new hisystologology videos. So I know this was useful for pathology residents who needed hisystologology refresher. H so there I am panning with all fours. Uh so uh I guess my

00:16:47
computer has enough GPU power to uh do smooth panning. What about data architecture and archiving? And if you have any questions, let me know. Uh also I will give you the link to the full paper. And if you want this presentation, let me know a presentation because I can give it to you. It's a PDF. uh I can just send it per email or upload maybe to LinkedIn just like as a carousel so that everybody gets access. So um comment presentation if you want the presentation and data architecture and

00:17:25
archiving um data stack uh that we are working with is daily intake uh for example 30 slides per day that's 300 uh sorry 300 slides per day 300 gigabytes and these are just examples right it's going to vary depending on the institution and we are taking this one gigabyte per slide uh as also an example. It can be less, it can be more. But let's say that was like the the average that they had. So uh one slide, one gigabyte. Then um we have so this is daily. This is every day, right? Then uh we have active archive 6

00:18:07
months to two years, right? Why two years? Uh that's what they site in the guidelines that you should um keep these digital images for. And it's going to differ by country, it's going to differ maybe by state, by regulation. Um, but because you need to keep glass also the like longer period uh legally required is going to be applying to glass whereas the digital is going to be probably the shorter period um for at least the active archive, right? or you have to um and I see you guys do want

00:18:46
presentation uh so ah and my upcoming publication uh talks about GLP certification for your system. Yes, GLP validation for primary read of um toxicologic pathology uh evaluation for drug development. And I mean I submitted the revisions. So I don't know yet when it's going to be out, but you know what? When it's going to be out, I'm going to put it in our digipath digest because hello, why should I not put my publication and my colleagues publication if it's about digital pathology? It totally fits. So

00:19:23
here active archive up to two years and then long-term reference um we have physical glass slides and retention uh depends how long 10 years for cytology 20 years for hisystologology that's what they are referencing um in the Polish guidelines it differs by country um but they are also working off of the um RC path guidelines we're going to see it and uh if we're doing things in the cloud Then obviously it um needs to be compliant. GDPR compliant in uh the uh European Union TLS encrypted connections. Why is

00:20:03
it flickering on me now? Apologies for the tech mandatory local expert capability. Why? Don't flicker at me. And they mention something um very interesting and important. No vendor lock in. I hope it's going to stop flickering cuz if it doesn't, I have to start drawing differently. Uh I don't know. Anyway, let's focus on stuff that works. No vendor locking uh is important because you may uh work with more vendors. consumer you may change vendors. So, uh this local expert is mandatory.

00:20:53
Um and I have couple of questions that are a little extensive. So, I'm going to get to them after we get through the presentation. So, stay stay tuned uh for question and answers at the end. And we said host light image 1 gigabyte. [gasps] Let me just I'm sorry I don't want to get the light of this flickering. Okay. So, compression and identification QC. Uh we have the original image and then we have a compressed image. Uh and they look the same to me on this picture. Uh and that's exactly the point. It has to

00:21:43
be the compression has to be visually lossless and and the compression standards mentioned in the guidelines are ratio 20 to1 visually lossless. So no diagnostic degradation uh all the diagnostic features are visible format or diccom these are open standards. Is that our reality? No it's not. But we are striving to have standardization in the file formats. So let's keep doing that. Um there are still a lot of proprietary formats but this uh vendor agnostic uh interoperability concepts they are

00:22:24
making its way strongly into digital pathology. So I think we're going to get there soon maybe this year. I don't know not everybody probably this year. than the identification of risk. Um the the fir the number one failure point is barcode readability. Uh and the the solution is to verify uh readability at different stages at the staining and scanning stages. And then uh important to have a QC protocol regular sideby-side comparison of compressed versus glass slides. um for the validation part, right? And that

00:23:09
takes us to the clinical validation philosophy. And I like very much this word [snorts] philosophy because um it's not one sizefits all for the validation. And I think I told you about that a couple of times. Um but so first of all the validation like prescriptions are going to differ depending on the regulatory framework you are um you are operating within right and the philosophy is that this validation is a pathologist process every reporting pathologist must undergo individual validation this is a model

00:23:53
that is um adopted in the uh RC path and it focuses on identifying risks in a real world workflow rather than abstract testing. And uh the goal and this is um that was an interesting um question that I got from uh my reviewers and um whoever like has done any kind of this work knows that non-inferiority is like a statistical concept for evaluating these scanners and comparing these two modalities. So, I got questions about non-inferiority in my publication, but uh basically what it means is the

00:24:38
digital diagnosis must be as accurate as light microscopy. It's not supposed to be worse. It's not going to be better because we're basically recreating the glass light. So, it's not going to be superiority, but it's going to be a non-inferiority. And scope is individual. Validation is specific to the person and the tissue type. So breast versus GI and different specialties are going to have different requirements. And I like this one very much. Let me put the smiley. Uh on identifying risks in a real world world

00:25:10
workflow rather than abstract testing. That's important. So how do they do it or what do they recommend? They recommend uh phase one which is familiarization and retrospective review the sandbox phase. So um they select a data set in this case 20 archived cases and mix of small biopsies resections stains uh H& IHC you can do special stains difficult features dysplasia mitosis and this is like an acrosstheboard concept philosophy basically h you want to see on digital what you see on glass

00:25:54
right and um like these difficult features the concept of difficult features or critical features may be different per specialty. H so for example my specialty path is going to have a specific list that was published and also we surveyed our pathologists at work. What do they think is difficult and we took all these these cases we found cases that had these difficult features. We scan them and we prove we have proven yes, you can see it on digital, right? And then we have dual review. I think when it does this funkiness, it

00:26:31
started flickering as well. No, dual review. Uh we have on the computer and under the microscope. [snorts] Um and this is still the sandbox phrase, right? Digital reporting, simulator mode, and glass slide review. Yeah, it starts flickering when it when I annotate too much. And then we compare we compare findings and the goal is technical prof proficiency with viewer tools which is taking the pathologist less and less time because uh all the new pathologists coming into the workforce they are screen uh native they

00:27:07
are digital native and the tools are also becoming more and more user friendly. So basically uh the the learning curve is like you sit click a couple of times and then you know where to click and you are proficient with the tools but if you're not or if you're resistant then you need to work on that or you know not do digital pathology and then we have the second phase the prospective routine diagnostics. So we do digital first. Pathologist formulates diagnosis on the computer screen and

00:27:40
signs out and state protocol signed by lab head authorizing independent digital reporting. And then we have also glass confirmation confirmation pathologist reviews physical slides. Uh and this so so this is like the the second part that they do for up to three months. Um, and I have seen a version of that in different institutions in different countries. In some countries, um, and I'm specifically talking about Japan. This, if you want to do digital, you have to do dual. There is no just digital in Japan. Um, and that's why I

00:28:22
admire the digital pathology trailblazers in Japan because they still have to do both. And then you document the discrepancies. Concordant, no change, a minor dis discordance clinically insignificant major discont discordance alters patient management. And these concepts are also uh described in um a paper published uh by rash scientists by Keith Wharton uh who is describing a validation of I think seven different scanners. Uh there's a podcast on that as well. Uh but basically he describes exactly how

00:29:03
did they validate scanners and these different validation approaches have the same concept embedded and it depends on where you are what kind of regulatory framework you're operating under which like bits and pieces but the concepts are the same and my favorite thing AI and by AI I mean um image analysis basically And now it's multimodel AI. So it's not uh restricted to image analysis. But um without digital slides this is not possible. And uh if you have hear me talking about uh the visual assessment of IHC scoring,

00:29:50
you know that my opinion is strongly against it. And if uh you know if there was an option to do it digitally I would do it every day of the week and twice in Sunday. Obviously visually is the second best and uh if we don't have the digital option then we do what we can but um the option to just even quantify IHC markers with artificial intelligence is something that is a gamecher and now we're entering an era where you're going to we're going to have uh digital diagnostics digital diagnostic tests

00:30:30
based off of image that are not visually accessible by a pathologist. Um some you can confirm but we are also working on molecular predictions of um molecular properties from H& uh supplement or replace molecular testing and this is only uh possible with computational pathology on images. So uh but for for the quantification this is enhancing reproducibility and not replacing judgment. There's always a pathologist in the loop. Um and the key applications that have been there kind of forever since we started doing image

00:31:17
analysis on digitized slides uh was the automated counting for Kai 67. her too. PDL1 is more recent but we have ERPR they are not that recent. We have tumor detection pre-screening lymph node metastasis. This is a super cool application that was also uh used for testing the performance of models in the chameleon challenge 2016 2017 and perfect application for AI and uh the AI tools must be validated against reference slides and they must prove non-inferiority to manual counting which I think is easier than um than

00:32:08
non-inferiority for these uh digital features the the difficult features h and tech stacks uh deep neural networks and convolutional neural networks but we have uh now the transformer architecture uh that is that has entered uh the scene recently we have foundation models so uh the the diagnostic tests are often still the ones mentioned here are often still uh based on convolutional neural networks working fine. Uh right, but we now are advancing on the computational um computational side as well. And um

00:32:50
yeah, soon enough there's going to be more. So there is also the evolution of intraoperative telepathology. So this intraoperative telepathology and this is I'm going to take my question here now because um there this question is related. So let's go through the slide and then I'm going to take the question. So um we have the site of the surgery and then we have a remote pathologist. This has been the application of digital pathology since its inception in remote Canada in remote

00:33:26
Scandinavia. Um but we also have here some performance metrics because uh if they are not met then there's like it's just so slow right so it has to take less than 5 minutes time from scanning to viewable image because hello we're doing a surgery time is crucial h and it has to be uh over 100 megabytes per second required upload speed um safety protocols microscopic supervision via live video link So the the pathologist needs to know what are they looking at and there needs to be redundancy backup

00:34:04
systems are mandatory surgery cannot stop right. So we cannot like oh go and take another piece of this tissue or like call me in half an hour. No. Uh and ah [clears throat] so now the question let's do the question um in my opinion is it possible one day to replace scanish with a special digital camera microscope which is less expensive and simpler to use. Um so replace scanners. Uh what I'm seeing is that you have different technologies in parallel all the time kind of always. Erh so it's probably

00:34:50
going to be an additional technology that is going to be used in places where this technology the scanners are not viable because of infrastructure because of cost because of uh I don't know internet bandwidth and there are already systems in place that enable um scanning from the microscope so with with a camera it's a kind of a manual process You can have also options um with um like this stage that is moving automatically. There's a word for that. Um motorized stage. Motorized stage. I was

00:35:31
losing looking for that one. Motorized stage. There's also an option to uh apply algorithms digital pathology algorithms on images. So there are workflows where the pathologist is doing the diagnostics on glass slides. But there is a microscope, there is a camera and you can take an image and this image can be um analyzed with an algorithm and you know the the Kai 67 can be counted. So it's like a hybrid workflow and there are also technologies being developed uh for scanning creating digital uh images directly from

00:36:07
tissue without making glass slides. So there are different uh things happening and different hybrid workflows. So scanner is one way of creating these digital images. Digital camera is another one. Smartphone is another one. Right? The question is always okay, how do we uh stay compliant? How do we stay um like how do we stay non-inferior? And then second, how do we stay compliant? Uh but yes, like start digging, looking uh at the podcast, looking at publications. I have people talking about alternative options that um are

00:36:50
more economical or you know um optimized for for different setups. So the answer is yes. The short answer is yes. There are options and we also have a checklist um infrastructure certified scanners it can be C IVD in Europe FDA cleared uh if but there is an option to have digital pathology as a lab developed test as well right but regardless uh if it's going to be a clear device if it's going to be a a clear framework for LDTS You need to have certain specifications and uh the resolution needs to be enough. It

00:37:38
has to be uh 0.26 microns per pixel or higher. Monitors need to be calibrated. Storage needs to be scale uh scaled validations. Uh so in this particular models pathologist complete uh phase one then pathologist complete phase two where they do the dual run and then uh you review the discrepancy protocols and if you reached a certain level of nondiscrepancies uh which like most of them over 90 I don't know definitely over 90 but I don't know if over 95 then uh you're good to go and the QC procedures active

00:38:19
visual versus compresses checks barcode readability verified at all stages. So um this risk assessment is also an approach being taken uh in the GLP validation that we did you basically identify risks and figure out how can we mitigate them um and if you are looking at the you want to look at the full publication um it's guidelines for the adoption of digital pathology in clinical pathology units recommended by the Polish society of pathologist ologist, diagnostic pathology. Uh, and they say digital pathology is not merely

00:39:02
a change of medium, but a shift toward greater precision, efficiency, and integration in modern healthcare, which I truly believe this is the case. And you know, if you're just if you're looking for what the specs should be, how you should do it, this is a cool publication that you can use. Very recent. Um, if you would like this uh presentation, let me know in the chat. Uh, if you're here live, let me know. Presentation in the chat. Just comment on whichever platform you are viewing.

00:39:38
I And now it's time for the gift. The gift. Well, this is not a gift. This is an offer that I'm going to be showing you. The Let's do the gift first, which is the book. The book is a gift for you because the PDF is entirely free. Uh the paper version is available on Amazon. So if you want to, you know, highlight and have the physical copy, totally available on Amazon. But if you want a searchable PDF at no cost other than me sending you emails, that is the cost. Uh then you can uh scan the code now. And

00:40:19
uh but so uh the thing that I prepared for you today is audio bundle. It's an audio bundle uh also created by notebook LM of the seven part AI in pathology series published by Rashid at alal in modern pathology. So uh I also have this uh uh pathology AI makeover course. I'm putting both of these codes in the um on the screen right now. In the corner you're going to see the course. So the course is also based on these publications but it's long. It's good. It's like you're going

00:41:02
to learn really well. But I realized that not everybody has the bandwidth to go through like a 5h hour course and um or I don't know it's probably longer even but uh they would still like to learn the concept. So what I did I created the audio bundle bundle AI uh powered audio bundle where you can listen to all these to not the publications because there's another tool I use for that. I'm going to tell you about that for like basically listening to publications as I do other stuff.

00:41:35
But uh these are podcasts created by notebook LM that are based on these publications with a specific prompt that uh like tells notebook LM who is listening to this podcast. So I described you precisely. So if you want to get the uh information from that uh pathology AI in pathology audio bundle er in under two hours that you can uh it's pretty like in an entertaining fashion pretty um like well explained without dumbing it down. This is your resource. It's a $19 resource. By the way, have you seen the new

00:42:20
Trailblazer mug? Some Trailblazers already have it. >> And there's going to be more of that because um I did publish this on social media already and a few of you uh got the bundle. So I am looking forward to your opinions. Let me know what you think because um what's happening we have these um live streams, right? And usually we are discussing abstracts. So, we have like five abstracts per live stream and we just look at the abstract and we think, "Okay, fantastic." Uh,

00:42:53
it's a good paper. Should I read it? Yes. Yes, definitely I should read it. Do I ever freaking read it? No. Uh, because life happens and you have to do stuff. So, um, but kind of like the success in a discipline like that that is very fast moving depends on us reading these papers. So, this is my shortcut. It doesn't replace the reading but my shortcut to learning a bit more than what we do in the abstracts in digipath digest. Digipath digest stays and we're going to keep uh meeting every

00:43:27
Friday almost every Friday. Uh but there's I am going to be creating these AI powered podcasts from different publications and I'm going to tell you about that next time for today. That's it. Uh if this AI impathology audio bundle is useful, let me know. Uh I listened uh through it all and it was very useful to me as a refresher. Even though I we read all the publications on digipath digest. They are included in the pathology AI makeover course. So I already knew it when I listened to this. I um got a cool

00:44:08
refresher to basically uh stay on top of things. Um, next week we're just going back to our normal uh digipath digest and I'm going to be announcing some new plans. Let me know if you want this presentation that we had today. Let me know if you're interested in the uh pathology around digital pathology around the world podcast series. And I talk to you on the next episode.