AI-Powered QC Where You Need It: Finding Needles in the Haystack

Agenda

Our agenda today will be to introduce Leica Biosystems, to talk about histological and digital artifacts in digital pathology, then introduce the Aperio iQC software, followed by a live demonstration, and hopefully have time for questions and answers at the end.

An Introduction to Leica Biosystems

An introduction to Leica Biosystems. The company has been in existence for over 150 years, with a focus on continual innovation in cancer diagnostics.

Our mission is advancing cancer diagnostics and improving lives. Our vision is from translational research to patient treatment. We envision a world where comprehensive insights enrich cancer diagnosis and offer reassurance to all patients.

Today, Leica Biosystems can enable more than one million cancer diagnoses globally every three days.

Deep Pathology Workflow and Process Expertise

Leica Biosystems has deep pathology workflow and process expertise. Today, we’re going to talk about artifacts that appear within the digital pathology workflow. However, these artifacts can be introduced very early on in the workflow.

Leica Biosystems has products and innovation at each of these steps, from specimen handling, tissue processing, embedding, sectioning, H&E staining, IHC staining, and then finally, digital pathology.

But really, the creation of digital-ready slides starts so much earlier in the process. In order to have true use of digital pathology and downstream AI development, we really have to have the best high-quality whole slide images built in at every step of the process.

Comprehensive, Integrated Digital Pathology Ecosystem

Leica Biosystems has a comprehensive and integrated digital pathology ecosystem, and we divide this up into scan, manage, and analyze.

When we think about scan, moving from left to right, we have systems with lower throughput, moving into fluorescence. More recently, we have our Aperio GT Elite for brightfield high throughput, the Aperio GT 450 system for diagnostic and research use, and our Aperio GT 180.

At the end of this, you’ll see that we have the Aperio iQC system. This is the software that we’re going to talk about today, but it really is part of the hardware solution because the purpose is to find artifacts and issues with imaging as early as possible in the digital pathology workflow.

Leica also provides solutions for manage and analyzing. Our manage portfolio, our IMS software, is with a partnership with Indica Labs from HALO Link, Aperio HALO AP, and Aperio HALO AP DX.

Within our Aperio AI Store, our marketplace, we’ve got the technology open to multiple vendors and third-party applications for the ability to analyze these whole slide images for various different applications.

Customers Talk, We Listen

As part of the journey of creating a new product, the first thing we do in Leica Biosystems is we go and talk to our customers. We wanted to understand what was either the barrier to the adoption of digital pathology or what challenges our customers were having once they had adopted digital pathology.

What we heard is not what we were expecting. Our expectation was that there would be much talk around quantification of biomarkers and assistance in cancer detection.

But in reality, the recurring theme that we heard was the challenges our customers have when trying to quality control their slides before they are sent to a pathologist. How much time this takes, and how much overhead that takes.

That started us down the road to try and fully understand the impact of artifacts and their presence in digital pathology slides.

I’d like to hand over to one of our customers, Professor Weis, to talk about this area.

Artifacts in Digital Pathology – Friend(s) or Foe(s)?

Good morning or afternoon. Thank you very much for the opportunity and kind introduction to say some words about artifacts in pathology or digital pathology, depending on your workflow. Maybe I’ll try to answer the question of whether these artifacts are friends or foes in the context of pathology.

Pathology and Artifacts: Is this an Issue?

What is pathology and artifacts all about? Usually in pathology, we get a small tissue sample, like the leaf shown here, and our job is to predict, based on that small sample, the entire patient, or tree in that example.

I’m sorry, but I’m a nerd. Two years ago, a colleague in Florence compared our job to Galadriel from Lord of the Rings. We get a leaf or a small tissue sample, and we have to predict the future of the patient, and we produce some kind of text, which is a little bit like oracle text. At least if you ask a clinical colleague, he or she usually has problems with our text.

Pathology and Artifacts: Pathology is all about artifacts!

The problem about artifacts in pathology, or pathology and artifacts in general, is that we are analyzing small parts of a patient, but not in its original condition. To stay with the example of the leaf, it’s not green. We stain it with our blue and purple stainings, and that’s not the normal color of a leaf.

So it’s already an artifact we need for our diagnosis, but it’s a helpful artifact. Without that, we won’t be able to make our diagnosis.

But besides these helpful artifacts, there are also non-helpful ones, even dangerous ones. Dangerous in regard to the diagnosis at the end. So it can endanger the diagnosis.

To stick with Lord of the Rings, that mirror is also dangerous, to keep it somehow in that relation.

Pathology and Artifacts: The Pathologist’s view?

Artifacts in pathology are a little bit like that ball from Lord of the Rings, where you can see very important information.

There are some nightmares associated with artifacts. One nightmare usually is missing tissue. Imagine you have a case where only one small part of the tissue contains a tumor, and that one part is lost. You have no chance to make the correct diagnosis.

Also, a somehow similar nightmare is when you have additional tissue on your slide. For instance, you have a colon biopsy. If the colon biopsy is from adenoma and there is no adenoma tissue on it, you have missing tissue. If it’s from a patient without adenoma, but the adenoma from a different patient is added, then you have a false positive diagnosis.

Also, one pathologist’s nightmare is wrong staining. For instance, CD3 instead of CD20. It could lead to a T-cell lymphoma instead of a B-cell lymphoma. These are really horrors for a pathologist.

Another problem with these artifacts is that they can happen, but they have a very low frequency. For me, it’s a little bit similar to that black swan idea from Mr. Taleb. It has a very low frequency, but it has a high impact.

Actually, these artifacts, at least the bad ones, are black swan events, in contrast to the conventional artifacts which are needed to even make the diagnosis.

Pathology and Artifacts: The Good, the Bad, and the Ugly?

That was the idea of grouping artifacts into categories in regard to the pathologist. You can also group them in regard to your workflow.

During slide generation, artifacts can happen. If you have a digital workflow, it would be pre-scanning artifacts. You have your very nice image, and there can be, for instance, rifts in the image, missing parts, folds, and uneven tissue.

There can be staining artifacts, like uneven stain or more faint stain if you compare that one to the first one. There can be covering artifacts, like these very nice bubbles here.

These are somehow not very nice. So if you look at the movie title I cited above, they are maybe ugly, but are they really bad or dangerous?

If you look at these rifts and missing parts, these can be dangerous. Then you would be in the category of missing tissue again. If you have some staining-related artifacts, you would be in the category of wrong staining, which can also lead to wrong conclusions at the end.

If you have a digital workflow, you add an additional step to your workflow, and this additional step can have its own artifacts and errors. There are, for instance, unsharp areas, again missing areas, and these well-known stitching artifacts, which produce these nice stripes on your image.

Again, the question is, are these dangerous? Yes, they can be dangerous, and most of them can be grouped into the category of missing tissue. So if your unsharp area is that one single area with adenocarcinoma in it, you have no chance, actually.

Your last step would be slide or file reading, depending on what kind of workflow you have. You can also call it post-scanning artifacts.

There could be things like wrong file naming. For instance, you have two cases of prostate biopsies, one case without cancer and the other one with cancer, and either the slides or the files are somehow mixed up. At the end, you would have two patients with a diagnosis of adenocarcinoma, but only one patient really has one. That is some kind of nightmare.

Other problems arising in the post-scanning area would be file compression and file conversion artifacts, which lead, again, to image distortions like that gray bar here.

You can group them again into the category of missing tissue or wrong staining. It depends a little bit on how much of your file is compromised.

That’s also a hideous thing about that. There can be artifacts which cover the entire slide. They are very easy to recognize. But there can be artifacts only covering one part of your slide.

Pathology and Artifacts: The Pathologist’s Summary

To come to a summary, there are many wanted artifacts, but there are also many unwanted artifacts which are possible.

They can be grouped according to their position in the workflow, which is somehow academic, but they can also be grouped in relation to whether they are annoying or even dangerous.

At the end, the question is, what are the consequences of that? The worst consequence could be that they are not only annoying, but even dangerous for a patient in the diagnosis.

For me, as a pathologist, at the end, the question would be, what can be done about it?

To come a last time back to Lord of the Rings, here in that image is Galadriel fighting Sauron, and Sauron is a great deceiver. That’s his official or unofficial title.

What can I do as a small pathologist? I’m not Galadriel, as you can see.

I need to be aware of artifacts. I have to think about it every day, actually, in every case. There is some kind of mental load. I have to keep it in mind every time.

The problem is the frequency of these is very low. If I would have an artifact, or a bad one, every tenth slide, it would be easy to be watchful. But if you have only every hundredth or ten-thousandth case such an issue, it’s very hard to remember it at the end.

We have to check it every time, and we have to think about some security measurements. In the end, the problem is also, if I have the idea there could be an artifact, I need to do something about it. For instance, I need to do a rescan, restaining, resectioning, whatever.

Everything with a “re-” at the beginning of the name includes more time and more money at the end. So it’s not only about my mental load and my problems. It is, at the end, really a problem about time and money, which needs to be put into such tedious tasks like finding a needle in a haystack.

With that needle-in-a-haystack issue, I give back to Catherine, who has some kind of solution.

What Artifacts Are Most Important to You When Checking WSIs?

Thank you, Professor Weis, for providing the pathologist’s perspective.

Once we understood that this was a problem worth solving, Leica Biosystems went about doing some research to try and understand which artifacts we should target.

As part of this, we did some voice-of-customer work where we surveyed 75 digital pathology customers. We wanted to understand, from their point of view, what artifacts are most important to them when they’re checking their whole slide images (WSIs).

We divided the results into gray here, which is the digital pathology artifacts, and red, which is the histological artifacts.

We saw some trends starting to emerge. People perceived the most important artifacts to be missing and clipped tissue, out of focus, and image striping.

From a histological perspective, we saw things like air bubbles, folds or wrinkles, dirt and debris, and excessive mounting media. These are all pain points that very clearly are highlighted across these 75 customers surveyed.

We also wanted to try and understand the behavior of what quality control manually looks like for whole slide images within routine pathology labs.

Magnitude of Quality Control Challenges in Digital Pathology

We saw that the percentage of whole slide images that are QC’d before being sent to pathologists ranges between 1% and 50%. So almost up to half of all slides are QC’d before they go to pathologists.

Of the slides that are QC’d, our survey showed that 98% of them are QC’d within one hour of digitization. The quality control checks are very much in line with when the slides are being digitized. Most of them are happening within 60 minutes of digitization.

We also asked the customers, “What benefits would you see from having an automated QC system?”

The results came back that people were interested in quality metrics, improving turnaround time, ultimately trying to replace manual QC steps, trying to reduce the amount of slides that should be manually checked, reducing the amount of staff required to operate the scanners, and freeing up full-time employees for more important tasks.

University of Heidelberg Study to Understand Real World Artifact Occurrence

We wanted to try and understand the rate of real-world artifact occurrence in a laboratory. So we worked with the University of Heidelberg and digitized 1,008 whole slide images on an Aperio GT 450 scanner.

The cohort included H&Es and IHCs, and over 28 different tissue types.

What this data showed us was that around 19.44% had no artifacts present at all. The rest had some level of artifacts. The majority of cases had one artifact present and, to a lesser extent, two artifacts present. This gave us an indication of the rate of occurrence.

We then went in and manually reviewed these slides to try and understand what artifacts were occurring.

Within the digital space, we see things like missing and clipped tissue, out of focus, and image striping. From a histological point of view, there was a variance of all types of artifacts, including dust, folds, scores, chatter, tears, air bubbles, pen marks, and knife lines. So quite a range of artifacts were present. Again, this gave us a good indication of what real-world artifact occurrence looks like.

Accelerating QC in Digital Pathology with Aperio iQC

We wanted to evaluate what benefits Aperio iQC had in the real world when used for artifact detection.

In conjunction with the University of Heidelberg, we performed a study where 200 whole slide images were reviewed manually and then reviewed again with Aperio iQC. We observed three things.

The first was that the number of slides reviewed per hour can be increased by 2.5X faster when using the automated assist of Aperio iQC.

The second thing we found was that technician review time greatly decreased when using Aperio iQC. There was a 69% time reduction in finding artifacts when using the Aperio iQC software.

Finally, the number of artifacts that were actually detected increased when using Aperio iQC. Twenty-four percent more artifacts were found when using the automated assist.

Challenges with Manual QC in Digital Pathology

There are many challenges with performing manual QC in digital pathology. The first is the time it takes to review the slides.

In a laboratory that has around 1,000 to 1,500 slides per day, our calculations show that about 25 hours are spent on QC’ing the slides before they are sent to a pathologist.

We also learned that between 61 and 90 seconds is the typical time taken to QC a slide manually. When we look at our calculations based on histotech or technician time in the U.S., that works out at a cost of around $1 per slide to manually QC a slide.

Obviously, with anything that is manual, there is the chance to have subjective and inconsistent data.

Aperio iQC Automatically Detects Six Common Artifacts

With Aperio iQC, we’ve created a software application powered by AI that will automatically detect six common artifacts.

Firstly, out of focus within a whole slide image, image striping, missing tissue, and clipped tissue. These four are our digital artifacts. We also have pen marks and air bubbles for histological identification.

Why use Leica Biosystems’ Aperio iQC Software?

So why use Leica Biosystems Aperio iQC?

We know from talking to our customers that people have created their own AI tools to do this, or they’re perhaps using other vendors’ AI tools. There are three reasons, really, which make Aperio iQC quite unique.

The first is AI is occurring at the point of digitization. We know the throughput of our scanners. We know how fast they digitize, and we created the application to make sure it can keep up with the demands of the scanners.

We know that our average AI results can be returned within 31 seconds from point of scanning. So you’re getting the information about artifacts at the point of scan, at the right time to be able to do something about it.

The second thing is Aperio iQC is fully integrated into the scanner, so it’s optimized for our GT platforms. As you can see here, within the console of our scanners, you actually get the alerts from Aperio iQC to let you know when there’s an artifact present. Again, the advantage here is that the slides are still in the scanner. It’s still at a time when you can actively do something about it.

Thirdly, with all things within Leica, we’re highly focused on quality. Each of these applications has been optimized for our GT platform; so our Aperio GT 180, Aperio GT Elite, and Aperio GT 450. All of the images and optimizations are to ensure that the algorithms are very robust on these platforms. If any change or modification is made at the scanner, it is obviously filtered down into Aperio iQC as part of the development.

How Does Aperio iQC Software Perform?

Here we’re showing the performance of Aperio iQC. We have six artifacts that we identify with five different models. As you can see from the results, there is high-performing accuracy, with all cases above 93%. Sensitivity and specificity are also very high, which really shows how robust these applications are.

These results are from our validation of the actual algorithms, and we are talking in the thousands of slides that were used for this study comparison.

How Does Auto-Rescan Fix Scan Issues Before They Become Problems?

One of the greatest advantages of Aperio iQC is the ability to auto-rescan, and the objective here is to fix scan issues before they become problems.

What this involves is AI is run to try and identify any problems. If a problem is identified, the information is sent back to the scanner, and the scanner rescans the slide with new settings and therefore generates a new whole slide image. All of this is done within seconds. The idea is that when a person sits down to look at the slides, the problem has already been corrected with a new rescan image.

Aperio iQC Software Awards and Achievements

I’d like to highlight some of the awards and achievements associated with the Aperio iQC software. The first is that we optimized and trained across two stain types, both H&E and IHC.

Within IHC, we optimized across 33 different antibodies, including membrane, nuclear, and cytoplasmic. The whole slide images were generated from three different scanner types, so our Aperio GT series: GT Elite, GT 180, and GT 450.

We also worked with five pathologists to generate our ground truth. As mentioned, there are six artifacts identified within Aperio iQC. Nine patents were filed. It was optimized across 12 different tissue types, and in total, a little over 32,000 whole slide images were used to train, develop, verify, and validate this product.

It has also been the winner of the Danaher Excellence in Innovation Awards, and we presented our data from verification and validation at three different conferences: DPAI, USCAP, and PathVisions.

What Are the Benefits of Aperio iQC Software?

The benefits of Aperio iQC software very much focus on where intelligence meets quality. Firstly, we increase efficiency. Aperio iQC software reduces hands-on time for technician review by 69%. That’s freeing up pathologists’ and histotechs’ time for more valuable tasks.

The second is reduced costs. With our estimations, manual QC of slides costs around $1 per slide in technician time alone. Automate repetitive tasks and save money with the iQC software. Finally, better quality. Consistently using software and AI detects up to 24% more artifacts than those identified with histotechs.

Aperio iQC Software Product Demonstration

I would like to bring you over and give you a short demonstration of the Aperio iQC software.

Here we see the login for Aperio iQC software. I’m going to log in. If we think about the workflow, this is designed for somebody who’s going to be performing quality control checks.

When I log in to this system, I see that I have four scanners connected to Aperio iQC. These scanners could be any one of the three from the GT platform.

Very quickly, I visually see where we have issues. The intention of this product is to drive the user to areas where they can quickly navigate to what they want to see. I can see from my icons here that there are no artifacts on these three scanners, but this artifact has both histological and digital errors.

I’m going to click on this slide, and we’re now presented with a slide tray. From here, really, it’s very much dependent on the user and how they like to perform their QC. But let me show you some things to draw your eyes to.

Very quickly, visually, you can see that I’m in this particular scanner. I have four other scanners on display. There are color-coded icons for slides. In blue, they have histological issues, and in orange, they have digital artifacts.

All of these are searchable and filterable. I can search for slides that have no artifacts, or go back and work through all the slides in my slide tray.

There are other filters you can use. For example, this little icon shows me that I have never looked at this slide. I haven’t checked it yet. But I can filter for slides that I’ve already reviewed and accepted, or slides that I have reviewed and declined.

Again, how somebody navigates this page is very much how they like to work when they’re quality controlling slides.

Let me bring you straight into a slide and see what the workflow looks like. Here we can see, first off, some of the icons and tools I’d like to share with you. This is a viewing pane, very similar to other products, but this has been built to try and operate as fast as possible because it’s all about getting through the slides as fast as possible.

We can see here what scanner it’s attached to, the slide name, and where it is situated in the scanner, so the rack and the slot. You have your overview. You also have information coming from the GT 450, in this case, with green here being the slide area. So you can see very quickly if there is any missed or clipped tissue.

Over here, we can see the outputs from the AI tools themselves. We start off with digital artifacts. In this case, you can see that the options available to us are out of focus, missing and clipped tissue, or image striping.

All of these are grayed out, which means those artifacts have not been detected in this slide.

When I zoom out here, you can see histological artifacts. We have air bubbles and pen marks, and air bubbles is highlighted here in blue, which also corresponds to the region that has an air bubble.

Some users like to turn this off so they’re not influenced where to go, whereas other users very much like to see the region that has been identified.

Again, everything here has been optimized to try and get to the image as fast as possible. When you zoom in, the overlay disappears. That’s not an additional click. It just disappears so you can actually see the slide.

We also have some manual interaction here that I’m going to speak to, and obviously the thumbnail and the slide label coming from the scanner also.

Here we can see that there are air bubbles. AI has identified the air bubbles, but it is the end user who determines what happens next. If they like it, if they think it’s still acceptable, they can accept the slide and move on to the next, or they can reject the slide.

If I were to accept and move on to the next slide, you’ll see quite quickly here that the interface has changed. It’s now highlighted that there are, in fact, digital artifacts present. There’s both out of focus and image striping.

Again, you could turn on and off the overlay, but the majority of customers just go to the region to see if that is acceptable or not. The end user would pan around to see if they think this image is acceptable.

One thing that’s different in this slide, that you may have identified, is that as this is a digital artifact, there’s something that can be done about it. Here we see that this actually is the first scan. AI detected there was an issue and sent that information back to the scanner, and the scanner then digitized a whole new slide again.

In this case, the settings are configurable within the GT 450, but it has scanned the whole tissue area and rescanned the slide.

The advantage of this is now the histotech, or whomever is quality controlling here, can accept this slide and move on. They don’t have to go back to the scanner and re-digitize. That work has been done before they sat down to QC the slides.

I’d like to show you some other artifacts, or just another one that I think is pretty important, as we learned earlier, which is missing and clipped tissue.

This is an example of a slide that has been digitized. We can see in green the bounding box that the GT 450 identified as tissue. The slide was then run through Aperio iQC, and in orange and red here, we can see this is the tissue area that iQC identified as tissue.

Aperio iQC flagged this as an artifact. It’s missing and clipped tissue. It sent this information back to the scanner, and it has been rescanned.

Again, if you think back to the presentation we talked about, this occurs at the point of scanning. Within 31 seconds, AI is performed, and the slide is sent back to be rescanned.

Now, as the end user, when I look at this slide, I can simply select the rescan and see if it’s acceptable to me or not.

Here you can see this is a rescan image. It has all the tissue region. No other artifacts have been detected, but as the human, I get to QC it. Obviously, if I wish, I can go through and then accept or decline. Again, the navigation is all about speed of viewing, so you can keep just churning through slides, or at any point, you can go back to your slide tray.

Another artifact that I’d like to show here is pen marks, as we mentioned. In this histological issue, pen marks are present. As a customer or as a user, I can go and view the slide. Another workflow that is available here is, say, if I disagree and actually I perceive this slide to be out of focus and I want it to be rescanned. This interface allows you to select that this image is out of focus and reject the slide.

Once we reject the slide, information will be sent back to the GT platform, and the slide will be automatically rescanned. Again, it’s all about optimizing time and workflow, and getting the information to the human as fast as possible. But this loop allows us to go back to the scanner and rescan this slide.

Some other information I’d just like to highlight. Again, the slide tray is very much a visual indication of where to go. We have a lot of people who just like to work based on when the slides appear in the system. As I mentioned, some people like to filter and look at slides that are digital or histological. We have other icons here that also show you, for example, this one is telling me this slide has been sent back to the GT platform to be rescanned. Then I can see these icons show me what are the rescans. It’s very intuitive and very easy to see where I need to review and what I need to do.

As part of the workflow, we also want to be able to get this information out of the database. A user can select the download, and here in a CSV or Excel file, they will have a list of all the slides and what observations they made when reviewing. Here we can see if it has been manually accepted or rejected by a human, and the pending slides haven’t been viewed yet by a person. We have file names. We have the scanner it came from, the rack, and slot. We also have the free text that was typed in by a user. We have the status of whether it’s a rescan or from the original image, and we also have any artifacts as detected by iQC.

Again, the free text field is here, so a human can write in if they want it to be re-coverslipped or if they want more staining. All of this is exportable, and it can then be used as their own work list within the laboratory.

The one last thing I’d like to show here is the configuration. As an admin user, I have the ability to go in and adjust the artifact thresholds.

Depending on the artifact, you could adjust certain parameters around the percentages of coverage or the severity of the artifact. Where this is beneficial is if you’re trying to scan, say, archive tissue versus routine, you might have different benchmarks that are acceptable for you. This is available as an admin user within the software.

Thank you for your attention, and we leave it open now to some questions from the audience.

Aperio iQC Software Demonstration: Key Features

• Connected scanner dashboard: Shows connected Aperio GT platform scanners in one place, helping users quickly see where QC attention is needed.
• Color-coded artifact alerts: Uses visual icons to flag slides with digital or histological artifacts.
• Searchable and filterable slide tray: Lets users sort slides by artifact type, review status, accepted slides, rejected slides, or pending review.
• AI-powered artifact detection: Identifies common issues such as out of focus, image striping, missing or clipped tissue, air bubbles, and pen marks.
• Fast slide review tools: Provides a viewer built for quick navigation, with slide details, thumbnails, labels, and tissue overview in one place.
• User-controlled QC decisions: Allows users to accept, reject, or further review slides based on AI findings and their own judgment.
• Auto-rescan workflow: Sends certain digital artifact findings back to the scanner so slides can be rescanned before manual QC begins.
• Exportable QC records and configurable thresholds: Supports downloadable QC data and allows admins to adjust artifact thresholds for different slide types or lab needs.

Key Takeaways About Quality Control in Digital Pathology

1. Quality control is a major challenge in digital pathology. Manual review can be time-consuming, inconsistent, and difficult to scale as labs process high volumes of whole slide images.
2. Artifacts can affect diagnostic confidence. Issues such as missing tissue, out-of-focus regions, image striping, air bubbles, and pen marks can delay review or, in some cases, risk obscuring important diagnostic information.
3. Artifacts can appear at different points in the pathology workflow. Some arise during slide preparation, staining, coverslipping, scanning, or file handling, which makes early detection especially important.
4. Aperio iQC uses AI to help identify common artifacts at the point of scanning. The software detects six artifact types and helps direct users to slides that may need closer review.
5. AI-assisted QC can improve efficiency and consistency. In the studies discussed, Aperio iQC helped reduce technician review time, increase the number of slides reviewed per hour, and detect more artifacts than manual review alone.

Frequently Asked Questions: AI-Powered Quality Control in Digital Pathology

1. Why is quality control important in digital pathology?

Quality control helps ensure that whole slide images are suitable for pathologist review. If artifacts are missed, important tissue areas may be unclear, missing, or misleading, which can affect confidence in the review process.

2. What types of artifacts can occur in digital pathology?

The webinar discusses both digital and histological artifacts. Digital artifacts include out-of-focus areas, image striping, missing tissue, and clipped tissue. Histological artifacts include air bubbles, pen marks, folds, wrinkles, dust, debris, tears, chatter, and knife lines. See the table below:

3. Why is manual QC challenging for pathology labs?

Manual QC can take significant time, especially in labs processing hundreds or thousands of slides per day. It can also vary from person to person, which may lead to inconsistent review decisions.

4. What is Aperio iQC?

Aperio iQC is an AI-powered quality control software application from Leica Biosystems. It is designed to automatically detect common artifacts in whole slide images and help users focus on slides that may need attention.

5. Which artifacts does Aperio iQC detect?

Aperio iQC detects six common artifact types: out of focus, image striping, missing tissue, clipped tissue, pen marks, and air bubbles. The first four are digital artifacts, while pen marks and air bubbles are histological artifacts.

6. How does Aperio iQC support faster QC?

Aperio iQC identifies potential artifacts and presents alerts in the software interface, helping users navigate directly to areas that may need review. In the study discussed, use of Aperio iQC was associated with a 69% reduction in technician review time and 2.5X faster slide review.

7. What is auto-rescan, and why does it matter?

Auto-rescan allows scan-related issues to be identified and corrected while the slide is still in the scanner. If Aperio iQC detects certain digital artifacts, the scanner can rescan the slide with new settings, helping resolve issues before the user begins manual review.