Laboratory Quality Assurance and ISO Accreditation for Digital Pathology

Hello everybody and many thanks for joining us in this session of the digital pathology virtual summit. In this session we are going to have three speakers who join us from Leeds Teaching Hospitals NHS Trust and giving three presentations around the adoption of digital pathology and their experiences.

Our first speaker is Chloe Knowles, and she is a specialist biomedical scientist at Leeds and she will be bringing us through a presentation on laboratory quality assurance and ISO accreditation for digital pathology. Her talk will be followed by Dr. Bethany Williams, who is a digital pathology research fellow in Leeds Teaching Hospitals NHS Trust and the University of Leeds, and Dr. Bethany Williams will be bringing us through her presentation on maintaining laboratory and clinical quality with digital diagnosis. And then the third presentation in this session will be by Dr. Darren Treanor, who is a consultant pathologist and honorary clinical associate professor at Leeds, and he will be bringing us through the Northern Pathology Imaging Cooperative and talking to us about digital pathology and artificial intelligence in the NHS. 

We will be taking questions throughout using the questions functionality at the bottom of your screen. Please feel free to submit any questions throughout the presentations and then we will answer them at the end of the session. With that, I will hand over to Chloe to get us started. Thank you very much.

Hi everyone and thank you to the organizers for inviting me to present today. My name is Chloe Knowles. I'm from Leeds and I will be discussing laboratory quality assurance and ISO accreditation for digital pathology. I'll be discussing some crucial steps to setting up a digital pathology workflow in the laboratory and gaining ISO 15189 accreditation for digital pathology by sharing our experiences at Leeds. 

Disclaimer

A small disclaimer, I'm sharing and presenting our own experiences at Leeds of laboratory deployment and ISO accreditation that were both successful. Other institutions may have different approaches and experiences, but it's important that we all share these both positive and negative in order to help other colleagues go digital. 

About Chloe Knowles

A small bit about me. I studied biomedical sciences at the University of Bradford. In which time I took an a work placement year at the Histopathology Department in Leeds. I came back to Leeds in 2014 as a registered biomedical scientist and in 2017 I took a one year leadership and management fellowship in digital pathology, where I was also the laboratory lead for the Leeds digital pathology project. Currently I am a specialist biomedical scientist working on the Northern Pathology Imaging Cooperative to deploy digital pathology both at Leeds and across West Yorkshire. 

Digital Pathology at Leeds

Leeds is a fully subspecialized diagnostic department. And holds over 40 consultants, up to 30 trainees at any given time and over 40 laboratory staff. The lab there has been scanning 100% of our surgical histopathology workloads since September 2018 and that equates to about 1000 slides per day. That generates a terabyte of image data every single day. 

Going Digital - Laboratory

First, we'll be going through some of the most important things to consider when going digital in your laboratory and then the steps that we took at Leeds to achieve our 100% slide scanning milestone. 

Things to Consider

The main points I'll cover are stakeholder engagement, workflow, scanner location, slide preparation and finally going into a little bit more detail about the ISO 15189 accreditation process we followed at Leeds. 

Stakeholder Engagement

Deploying digital pathology will mean asking your colleagues to make significant changes to their roles. Engaging with them from the outset is essential for creating enthusiasm, but also for our successful deployment. Ask staff to provide for you feedback, express any concerns they have, or make any suggestions in terms of laboratory workflow. 

At Leeds we distributed a survey to all the lab staff to gauge how they felt about digital pathology and how they felt it would affect their day-to-day role in the lab. This allowed us to identify a diverse range of opinion and then target particular concerns or issues with focus groups or informal meetings. We also identified individuals who are particularly enthusiastic about digital pathology. They can help implement and sustain the changes in the laboratory and also encourage other people who are less enthusiastic. 

Workflow

In regard to workflow, it is very important to consider your entire laboratory workflow from receiving a specimen to reporting a case. Documenting your current laboratory workflow gives you a better understanding of the pressure points, bottlenecks, and potential waiting times that you have in your current process. Doing this also allows you to create a future workflow with digital pathology incorporated rather than added on as an additional step in your process. It also allows you to identify any waste that you may have and work out ways of eliminating it in your future state. 

Ways you can do this is using lean principles or using continuous improvement concepts. This is a great way of mapping out existing workflow and preparing your future state. 

Scanner Location

Next is scanner location and this is a crucial step in the deployment as it reaffirms your workflow plans. It is important to consider which locations in your department are best for introducing scanning into your workflow, and it's also important to take your time to consider where your scanners should be placed. The Leeds approach to digital pathology in the lab was that the scanners should be an integral part of our laboratory processes. 

We held a formal options appraisal of ten different locations in the laboratory and considered aspects such as walking time, hands on time and how well that location supported the current laboratory workflow. In the end, we chose a centralized bench adjacent to the main lab and quality control areas as this kept the scanning visible and part of the main laboratory workflow. 

Slide Preparation – Microtomy and Coverslipping

Quality of sections and slide preparation is already an integral and well-established part of the histopathology process without digital, but it becomes even more important when preparing your slides for scanning. Sections should be thin and free of bubbles, folds or creases, and the section should also be in the center of the slides where possible. There should be no overhanging or broken coverslips and the slide should always be free of excess mounting around the edges. 

Slide Preparation - Scanning

Another important aspect of slide preparation that you may need to consider when implementing digital in your laboratory is ensuring that the slides are dry before they are scanned. At Leeds, we incorporated a drain step and we found that ten minutes at 60° is a sufficient drain time. 

The slide should also be free from any dirt, pen or wax, and if cleaning is required it should be done with a microscope lens cloth and 70% alcohol if necessary. Checks for these kind of issues should be done during your routine laboratory quality control checks before they are sent to the scanner as it keeps the digital pathology at the forefront of people's minds when they're looking at the quality of their slides and it also makes the scanning process more efficient. 

Top Tips for Laboratory Deployment

Finally on workflow these are my top five tips for laboratory deployment based on the experiences we had at Leeds.

1. Involve all staff from the start. This includes laboratory staff, pathologists, management, local and corporate IT. Anyone you think will have a part to play in making your digital pathology deployment successful. Get them involved and get them involved from the very beginning.
2. Mapping out your current workflow will make it easier to identify any potential issues that you have in your current processes. They can be easily removed when you start to plan your future state for digital pathology.
3. Make slide scanning visible to ensure that it is seen and maintained as an integral part of your laboratory workflow, rather than being considered as an additional step.
4. Introduce quality control for scanning slides before they are sent to the scanners. Having these quality control checks as part of your routine laboratory checks contributes to the point above about making digital pathology integral rather than additional.
5. Remember that implementing digital pathology involves a great deal of change for a lot of people. It is essentially a change management process in the laboratory to deploy digital pathology, and it is essential that you bring your entire department with you.

ISO 15189 Accreditation

Moving on now to accreditation for digital pathology, and I'll be talking through our specific approach to gaining ISO accreditation for the digital pathology workflow at Leeds. Histopathology laboratories fall under ISO standard 15189-2012, which is medical laboratories requirements for quality and competence. In the United Kingdom, Laboratories are assessed by the United Kingdom Accreditation Service, which is comparable to CLIA in the United States. 

Accreditation in the Laboratory

The Leeds approach to gaining accreditation in the laboratory covered the following aspects:

• Change control risk assessment
• Verification and acceptance
• Comparability and reproducibility
• Training and competence
• Uncertainty of measurement

I'll go through and talk about these individually. 

Change Control

Change control is a vital part of digital pathology implementation, both in terms of deployment and gaining accreditation. Having a change control procedure in place ensures that all changes made are documented and appropriately managed. It also, and probably most importantly, ensures that disruptions to existing clinical services are kept to a minimum. As digital pathology was a new process for Leeds, we developed and adhered to a full change control procedure. This allowed us to identify key people, manage the collation and submission of evidence to UCAS, and to ensure that all aspects of the deployment were covered and controlled. 

The change control also has benefits if findings are raised during your inspection. And it can simplify the process if it's implemented correctly. Examples of key people that you may want to include when starting your change control procedure are people like the clinical lead for the department, any health and safety officers to help you write risk assessments, a laboratory lead who understands the workflow, and a quality manager who can help with all the gathering of evidence and submitting paperwork. 

Risk Assessment

Any risks in terms of health and safety for your digital deployment must be identified and mitigated before it can go live. At this stage it's crucial to involve a health and safety representative who can assist you in completing any health and safety documentation, and in particular the risk assessments that are required.

The type of assessments that you would expect to complete are things like the equipment risk assessments for any scanners and workstations that you intend to use. The proposed workflow for scanning to identify any risks in that process and how they can be mitigated if necessary. A general risk assessment to include the environment in which the scanners will be placed and where any staff may be working to ensure that they are safe. And a display screen assessment for all newly trained staff should be completed to ensure that they're comfortable working at the workstation in the laboratory or for the pathologist at their workstations in their office. 

Verification and Acceptance

For ISO 15189 accreditation, laboratories have to provide evidence that the scanners and any associated software have been tested for their intended use and are working as the manufacturer states and as the laboratory expected to work. This must be in the form of a written document and it must be submitted to demonstrate that these tests have been carried out. 

One way of providing this documentation is to perform tests against the original installation checklist that was provided by the manufacturer when the scanners or any software was originally installed. 

Comparability and Reproducibility

If multiple scanners are going to be used as part of your digital pathology workflow, then evidence should be submitted to demonstrate that all the scanners you use produce images of comparable and reproducible diagnostic image quality. This can be achieved by scanning the same set of slides, typically H&E and immunohistochemistry stains across the scanners, and having the images assessed by an experienced and validated digital pathologist. The features of the images you can assess include things like image focus and staining quality. 

Laboratories can also use interlaboratory assessment schemes, which would involve other laboratories scanning the slides on all scanners available in their department and having the images assessed at different sites. If you can't take part in an interlaboratory scheme, then another way of providing evidence for this is to rescan a previously digitally reported case and asking the same pathologist to revisit, just to check that the images are the same and the same diagnosis will have been reached. 

Training and Competency

Training and competency is a vital part of both laboratory deployment and accreditation. It ensures that the level and the different types of training given are specific and consistent. It also demonstrates a level of competency and identifies areas for staff that they can improve on. 

In order for training to be efficient, the standard operating procedures for all the tasks that the lab staff are expected to carry out on a day-to-day basis should be available. Any user manuals for the equipment and software that they will be using need to be available at all time. 

Alongside the manuals and standard operating procedures, any training booklets and competency framework assessments should be completed to demonstrate that the staff's understanding of the digital pathology workflow and associated tasks. These standard operating procedures and training documentation will require updates whenever changes are made to the processes or workflow and they should also be version controlled. 

Measurement of Uncertainty

If you are using digital measurement software, then uncertainty of measurement will need to be considered. A calibration slide should be utilized to assess whether objects in images have been captured and that they are to scale and correct. It is also important that the manufacturer of the slide, the calibration slide, provides a certificate that includes data and demonstrates that the slide is fit for purpose. 

As a minimum when measuring when checking for measurement of uncertainty, a width and height measurement should be taken, and it should be recorded to monitor any changes in that measurement. A reference range should also be provided with the calibration data on your slide so that you can ensure that the height and width measurements are still in range.

This should be done across all your scanners to be used in your digital pathology workflow. To demonstrate that the measurements taken are safe for diagnostic use, an audit should be carried out. This ensures that any clinically relevant measurements are valid. This can be done by asking pathologists to measure the same area of a slide. For example, a tumor measurement or a margin distance once a day for one to two weeks, enabling any variation in measurement to be documented. This is a crucial part of 15189 accreditation.

The assessors like to see that you've demonstrated that you're checking these measurements and you're maintaining within the reference ranges.

ISO 15189 in Numbers

This slide just demonstrates the ISO 15189 accreditation process for Leeds’s numbers. There were a total of 14 people involved, 62 documents were submitted in total, both technical and clinical. This equated to around 6000 hours of time spent creating, submitting, and checking all the evidence. 

After our first inspection, there were seven findings and finally, in October 2018, one accreditation for digital pathology under 15189 was granted by UCAS. The accreditation process is challenging and as you can see it requires a lot of work by a lot of people. But knowing that our digital pathology workflow is now being deemed safe for both patients and staff, it's an important achievement when using digital pathology in a clinical laboratory. 

Top Tips for Digital Pathology Accreditation

To summarize, and from Leeds experience, these are the top tips for digital pathology laboratory accreditation. 

• Begin with change control so this process and documentation is key to both the successful deployment and accreditation process.
• Complete the risk assessments for all aspects of the process and this keeps all members of staff and patients involved safe.
• Document scanner performance against the manufacturer. This is an essential part of ISO 15189 accreditation and must be submitted as a written document with results.
• Ensure that all scanners used produce the same quality of image as this promotes comparability and reproducibility of diagnostic image quality.
• And finally, conduct an audit for uncertainty of measurement, and this is vital if the images are being used to report diagnostic cases. 

Get Your Free Downloads

Everything I've spoken about today is available for download from the Leeds Virtual Pathology website. The Leeds Guide to Digital Pathology gives details on laboratory workflow and deployment, as well as any clinical aspects. The ISO 15189 practical guide paper that is published in the Journal of Clinical Pathology is also available to download. 

Acknowledgements

That leaves me to thank the organizers again for inviting me to speak today. I'd also like to thank the histopathology department at Leeds and the whole Northern Pathology Imaging cooperative team. I hope you found the information I've given here today useful. Thank you.

Maintaining Quality Diagnosis with Digital Pathology

Hi, I'm Dr. Bethany Williams and first I'd just like to say how excited I am to be joining you at Leica Biosystems Virtual Digital Pathology Summit today. Today I'm going to be talking about how we can maintain the quality of our diagnosis with digital pathology. We want to ensure that we as pathologists are positioned so that we can benefit from all the quality and efficiency advantages of whole slide imaging over conventional light microscopy without compromising our professional standards or patient safety. 

Overview

I'll start by introducing you to some of the work we've done on our 100% digitization at Leeds Teaching Hospitals NHS Trust. In the UK, then I'll speak a little bit about some of the clinical considerations to keep in mind when you are planning for your ISO inspection of your new digital pathology services. I'll describe a little about the clinical training and validation programs we've got in place for our pathologists at Leeds. Then I'm going to give some hints and tips on practical digital reporting for the novice digital pathologist. At the end I've got a little time to talk about remote reporting, which has become pertinent in light of the COVID-19 outbreak. 

About Leeds

Firstly, welcome to Leeds. Leeds is situated in the north of England, Leeds Teaching Hospitals NHS Trust is a single site histology laboratory with a fully subspecialized diagnostic department. We've got in the region of 45 consultant pathologists supported by 30 trainee pathologists. 

We see something in the region of 300,000 H&E Histology slides per year and this equates to about 1% of the total NHS histopathology workload. We're very lucky to have a digital pathology team consisting of doctors, laboratory staff, university fellows, and project managers, and headed up by Dr. Darren Treanor, who is the Royal College lead for digital pathology and we're very proud that we successfully passed our ISO inspection of our digital pathology services in recent years. 

Clinical Usage

At the moment, 100% of all our histology slides are digitized and available for diagnosis. To enable this, we've had to install and deploy six high-capacity scanners and three smaller scanners which we use to do our mega block scans and to add little bit of extra urgent capacity. 

We aim to do continuous small batch scanning throughout the day, particularly over urgent biopsies and small specimens. Then there's overnight scanning of resection work. Now, in order to implement this, obviously we've had to devise a lot of new workflows and a lot of new roles in the laboratory. 

Now looking at our diagnosticians, so all our breast and neuropathology diagnosis are now made on digital slides. With a proportion of gynaepathology, renal, GI, liver and lung specimens also reported digitally. All our pathologists are either validated to make primary diagnosis on digital slides or are in the process of an individual validation scheme. I'll describe that in a little more detail later. We're all using medical grade eight megapixel diagnostic displays in the department.

Deployment Rollout in Leeds

What was our road map to 100% digitization? When we started out in 2017 with a pilot project to digitize our breast primary diagnosis. It was very swiftly followed by neuropathology, primary diagnosis and frozen section. Then in 2018, we did a standalone immunohistochemistry deployment for all those subspecialties that weren't having their primary histology scanned. We also ramped up scanning in the laboratory until we reached that great 100% digitization milestone. 

Since 2019, we've had all our pathologists at work validating their digital practice in cohorts and we've also commenced some exciting regional networking projects. These are part of the Northern Pathology Imaging Collaborative that Dr. Darren Treanor is speaking out about in the accompanying presentation at this summit. 

ISO Accreditation for Primary Digital Diagnosis – How Did We Do It?

Let's start out with a few words on ISO accreditation for primary digital diagnosis. One of the questions I'm asked most often is sort of how we managed to achieve this. Obviously with digital pathology being a very novel process in the laboratory, there's been a lot of concern about how we can successfully pass an inspection. 

Can Your Pathologists…

When you're preparing for your ISO inspection. You really want to ensure that you've reached a point where you're confident your pathologists will be able to demonstrate and explain how they report a case using their digital pathology software. You want to be happy that they can identify and report any artifacts or any other problems with the system. 

They should be able to explain their validation experience and their continuing professional development. Describe the scope of safe practice for digital sign up. Are there certain types of cases which they know they're not comfortable to sign out, and what they do in those circumstances and should they all be able to access supporting documentation? 

Supporting Documentation

This documentation will include generic documents such as standard operating procedures on how to report digital slides, how to validate your primary digital slide reads. They should have access to their software user guides and manuals. Also data collection templates for reporting exceptions or reporting any problems with the system. 

They should also have their own personalized folder, including things like their screen display risk assessment. Any of their validation or training certificates that show that they've completed a training procedure, a record of training validation meetings they've attended. And summary statistics for their validation. I'm going to speak a little bit more about what that validation process entails a little later. 

A Few Words On Inspection

In general, when I'm talking about inspection, I think it's important that this isn't just a tick box exercise. Preparing for an ISO inspection takes a lot of time and a lot of effort. It's worth making sure that all this time and effort adds value to your laboratory technicians and your diagnosticians’ daily work. Ensure that these procedures are meaningful and have real training and developmental value to your colleagues. They're not just a tick box exercise. 

Everybody Expects an Inspection

The one thing we can expect in laboratory medicine is that we are going to get an inspection. It's important once you start to know about the time scales for your inspection process that you identify the key team members who are going to be responsible for delivering different aspects of the inspection both in the laboratory and in the diagnostic office. And that you regularly review your progress and documentation as you head towards.

I've spoken quite a bit about validation and training. But also, very keen that this should be an evidence-based document. To make an evidence-based training program, we really wanted to work out what were the real potential pitfalls of digital diagnosis. What are the real training lessons we can learn from the established literature and the wealth of clinical experience that's available in those few key centers that were early adopters of digitization? 

Potential Pitfalls of Digital Diagnosis

I would refer you here to our paper a systematic analysis of discordant diagnosis in digital pathology. Now in this systematic review and analysis, we're able to identify key themes and areas where novice digital pathologists seem to have more issues when they first started reporting digital slides. 

These included dysplasia detection and grading, navigation in slides and the detection and location of small objects. There were certain sort of specialty specific textures and objects that people seem to repeatedly have some difficulty diagnosing. 

Diagnosis and Grading of Dysplasia

What we see in the literature is that there is a real tendency to either miss dysplastic lesions or to undergrade or under call these lesions. There are two things that are happening here. In some cases, pathologists were missing the appropriate regions of interest on their initial low power screening of the specimen. This occurs particularly in specimens like cervical biopsies or loop excision biopsies. 

There were also problems implicated in the high-power view of some of these dysplastic lesions. Here it was nuclear detail; rendering of fine nuclear detail that was causing problems. Luckily there's great potential to mitigate risk on your digital slide reads of dysplasia. 

Firstly, the most important thing we can do as pathologists is educate ourselves and there's a definite individual learning curve for successful dysplasia identification and classification. There should be a real focus of your early training as you transition to digital slide reporting. 

As you start reporting dysplasia here on digital slides, you can modify your workflows to introduce temporary safety checks. You could have mandatory glass slide checks before you sign out your digital Barrett’s esophagus cases, for instance. There is a definite case for routine scanning at 40X of all diagnostic biopsies just to maximize your potential to pick up some of those trickier dysplastic lesions on diagnostic specimens. 

Finding Small Diagnostic Objects

The next area where pathologists often have difficulty is locating small diagnostic or prognostic features or objects on those digital slides. These, from the literature, can be a range of neoplastic and inflammatory lesions. The thing that people most commonly have difficulty with is picking up small foci of inflammatory activity and granulomatous inflammation. There are also problems with certain microorganisms; things like your helicobacter and candida, which can be difficult on the glass slide reading as well. 

Quite crucially there are sort of neoplastic lesions. Things like the lymph node metastasis and micro metastasis, and tumor microsatellites, which historically we have seen in the literature, can be missed on digital slide reader. Please refer to that paper that I indicated if you want to find out more about those types of case. 

Now what's implicated here? Clearly it's navigation. Navigation within slides and navigation between slides. Luckily, again, there's a real potential to mitigate the risk of digital diagnosis. We just need to emphasize the importance of pathologists to training in the safe, efficient use of the viewing software. You can also make use of safety features like thumbnails and slide indicators which tell you which slides and which regions of slides you've viewed or not viewed when you've been looking at a case. 

You can incorporate checks into your workflow when you first start reporting digital cases. Perhaps you want to review your digital diagnosis of Sentinel lymph nodes, for instance, and just do a quick glance slide check before you sign those out. I would emphasize the importance of 40X equivalent scanning, particularly if you're going to pick up things like microorganisms on your stomach biopsies or pylori, or candida on your oral biopsies. 

RCPath Guidelines for Digital Implementation

With this in mind, we developed our training and validation protocol at Leeds, which now forms the basis of the Royal College of Pathologists guidelines for digital pathology implementation and validation. 

How Do We Train and Validate Pathologists?

This is a very novel validation protocol which was designed to be pathologist centric. It's an individual validation for an individual pathologist. It's based on the evidence base that I've described and it enables the pathologist to self-identify their own digital diagnostic pitfalls, gain confidence in a risk mitigation environment with early exposure to blind digital reporting. 

There's no need to review many archive material. You're training using your standard workload. This process has also enabled the pathologists to develop their own workflow modifications to really help them reach that stage where they're not just competent, but they're confident in their digital process.

Validation Summary

Validation for each pathologist starts with a one-on-one formalized training in the use of the digital microscopy system and crucially observed practice with feedback. Next, the pathologist is presented with a training set of up to twenty challenging and informative specialty specific cases, which are all relevant to that pathologists workload. They view these slides on the digital, make notes, make their diagnosis, and then immediately compare with glass to reconcile their reading of the digital slide with that of the glass slide and note if there are any differences in their appreciation of the case on the two media. 

Next, they move on to the live phase of their validation. Here the entire workload of that pathologist is scanned prospectively. We do we aim for roughly around a two-month, whole-time-equivalent workload, but this varies from pathologist to pathologist based on the type of work they're doing and how easy they find the transition. We ask to make all their diagnosis on digital slides, but with an immediate glass check before they make that final crucial sign out so that they can record if they have any difference in opinion once they've reviewed that glass slide. 

Breast Pathology Validation

We first used this on a cohort of breast pathologists at Leeds and there's the reference for you there. This was the part of our pilot study that I mentioned on the initial slides. We had three participants in the first cohort, a mixture of enthusiasts and skeptics. They viewed all these cases on 40X scans on six megapixel medical grade displays. Now in their training sets, over 20 challenging cases, they had difficulty counting mitotic figures, recognizing micrometastases, identifying weddellite calcification as an example just there in the bottom image, and recognizing and grading nuclear atypia. 

Live Reporting

In their live reporting phase between them, they saw just under 700 cases, 3 1/2 thousand slides, and a range of formats and stains. And as you can see, for our cohort of breast pathologists, there was an observed rate of complete clinical concordance between the glass and the digital slide reads of 98.8%.

Reporting Preference

We also asked our pathologists for every case they looked at. Did they prefer the glass slide presentation, the digital slide presentation or did they have no preference in reaching their diagnosis for that case? 

I always put this slide up to prove the point that if you ask three pathologists a question, you're going to get three very different answers. As you can see, one of our pathologists had a very clear preference for digital in the majority of cases. Another and others had a greater preference for glass, but all of our pathologists were in agreeance that for the majority of cases they either preferred digital or had no preference. All pathologists also agreed that for a certain proportion of cases there was a clear advantage or a clear preference to consulting the glass slides for that case. 

What Did They Think?

Following their training and validation, our pathologists are able to make 99% of their diagnosis on the digital microscope, deferring to glass for that 1% of cases that they still find tricky. And for our breast pathologists, these are borderline mitotic counts, which might affect the final grade of the tumor and cases where they failed to detect weddellite in a screening specimen. For pathologists at the moment, the light microscope is strictly for emergency use only. 

Practical Digital Reporting

Now on to a few sort of practical hints and tips about digital reporting. First of all, let's think about our pathologists. This is an image a lot of people use when they're talking about the attitudes of their colleagues towards digital pathology adoption and reporting digital slides. This is a very outdated image and it's very unfair. 

In our department we prefer to divide our pathologists into three groups, three teams. We've got our enthusiasts, our cheerleaders, enthusiastic people. They're great to bring into meetings when you're trying to sell your business case for digital adoption. We've got our skeptics who are the people who can point out all these little flaws and problems with all our digital pathology plans.

They're also useful people when you're planning your deployment, because they will see the problems and the potential pitfalls that your enthusiastic colleagues might be blind to. 

Then we've also got a group in the middle. We've got our uncertains, our agnostics who really aren't quite sure. We've got these different groups in our department cause we've all got very different attitudes to change, different attitudes to risks, and different starting level of IT competency.

Many of the people in the uncertain or agnostic group have sort of underlying fears and concerns, so it's important to seek early opportunities to engage, particularly with this group, and to make sure you optimize their initial experience of digital pathology and share early successes as feedback. 

Validation

Now our validation and training protocol, we think covers all bases. It allows our enthusiasts early live reporting experience. They can do digital. Our uncertains get that safety net where they're able to explore digital, but with those glass slides- slide safety checks. Our skeptics get a chance to put your digital system to the test. A chance to prove you wrong. 

Common Issues – Input Devices

What issues do we find our pathologists have when they start on their validation? A lot of their issues are very basic and do with their use of input devices. It's basic ergonomics. I see a lot of very crazy, exaggerated movements of the mouse. I see lots of claws clutching at the mouse. This is why it's important to observe your pathologist using the system and give them one-to-one feedback on how they can use their mouse more efficiently. How they can use shortcuts and if possible, you can give your players a choice of input devices, including gaming mice or more joystick type input devices. 

Common Issues – Navigation Within Slides

One of the biggest issues our pathologists have initially is navigating within slides. You've got your beautiful medical monitor, you've loaded up your first digital slide and you get this amazing, beautiful quality image of your central lymph node less than one times equivalent magnification. It's very easy to look at that and think, “well, job done, this is great, this is a clear lymph node.”
We call this phenomenon the whole slide seduction, because if you had just taken a few moments to increase your magnification to 5X, you would have easily spotted that sneaky micrometastasis there.

That initial full screen, very low power view of the digital slide which provides you that beautiful low power overview of architecture, which you just can't replicate on the live microscope, and can come back and bite you. You do really need to ensure that you still screen all your slides at a task appropriate magnification to avoid missing the crucial regions of interest. 

The hardest tasks initially are often to these high-volume tissue search tasks. In our experience, using a bright high contrast, high resolution medical grade screen usually helps, but you still just must remember to be as systematic and disciplined as you would be on glass. And of course, when we're first training on the light microscope, we learn our little lawn mower moves to make sure that we've seen all of the material presented on the glass slide. Obviously you need to train yourself how to how to replicate that on the digital slide and develop your own technique. 

Common Issues – Artefacts/Focus Issues

Other common issues you need to make sure your pathologists are aware of the types of artifacts you can get on digital slides. You’re going to get any artifact that was present on the glass slide replicated on the digital. You can also get new types of artifacts, including things like this striping demonstrated here. Any digital cases with striping should be reported to the lab scanner team so the slide can be rescanned for the archive. The pathologist needs to decide on a case-by-case basis when they're prepared to make a diagnosis on a case with a striped slide or not.

Common Issues - Special Stains

There are certain tinctorial stains that are quite hard to represent on the digital slide. This is a gomori stain here in a neuropathology case, a muscle case, and it may be that you need to adjust scan settings or laboratory staining protocols with some of those darker tinctorial stains. You just need to be familiar with how these stains appear on a digital slide.

Validation Scenarios

Most pathologists in our department all report a single or at most two different pathology subspecialties. Your training should be specialty specific. And your training sets should reflect your local specimen types, staining, special tests and immunos. You want to try and cover the full range of diagnoses and tasks an individual pathologists gets in their regular workload and make sure you're including enough pitfalls like some of those things that I pointed out earlier.

I said we've got this paper, the practical guide to training and validation for primary diagnosis, which could be very helpful in putting together training. 

Validation for Generalists

For general pathologists, the same principles apply. They should still get a training set to covering a number of different specialties and all the commonly encountered issues; dysplasia, lymph node searches. We'd recommend additional specialized training sets if you're the lead for a particular subspecialty. It just really needs to be cases that are relevant to individual pathologists. 

Generic “Difficult Things”

The generic, difficult things you'll want to cover, are diagnoses dependent on fine nuclear detail. Things like papillary features in thyroid cancers, things like selecting small regions of interest in large tissue dissections. Include things like micrometastases, detection of microorganisms, differentiation between granulocytes and some of those refractile eosinophilic items as well as those mitotic figures, which can be tricky to identify and count.

Digital Immunohistochemistry Assessment

20X scanning is sufficient for confident assessment of the vast majority of immunohistochemistry slides, but 40X really adds some extra value if you're trying to evaluate H Pylori, Ki-67, PDL1 Her2, and ER and PR. Any of those stains that need a little more refined approach to assessment.

Remote Reporting – Is it safe?

Finally some words on remote reporting. One of the questions I'm asked all the time, is remote reporting safe? The quick answer is it depends how you do it. There are all sorts of factors to consider. The type of case that you're reporting, the quality of the scan that you're reporting, the workstation that you're using, and in particular the screen, the pathologist that's sitting there and doing the reporting and the fail safe and support network that are built into that pathologist’s at-home network. 

In general, you want to think is that pathologist suitably trained and validated to work with those digital slides. Are they competent and confident in their diagnosis? Is their viewing hardware suitable for the purpose? Do they have access to support fail safes? Do they have the backup of being able to review glass or refer the case to a colleague? 

For Information Relevant to COVID-19

Now this has become particularly relevant recently considering the COVID-19 outbreak. There's been a lot of renewed interest in home reporting and there has been some temporary relaxation of the national and international guidance for home reporting of specimen. I would refer you to the Royal College of Pathologists’ COVID-19 resources hub where there is a new guideline issued by the Royal College and also to a paper in the Journal of Pathology Informatics, which we've wrote, which gives guidance on remote reporting of digital pathology slides during periods of exceptional service pressure. This includes access to a unique point of use quality assurance tool that you can use. It's free to access. It's a very quick little test you can use to see if your home computing monitor is suitable for reading H&E-stained histopathology slides. 

A Meaningful Validation Process Will Result In:

To summarize, a meaningful validation process is going to result in pathologists that are confident in their abilities and limitations with digital diagnosis, who are familiar with their hardware and software, and can recognize and report performance issues, a department which understands and has investment in its digital pathology system, and a department which could develop its own bespoke ways of using digital to improve outputs, workflows and working environments. 

I hope you found the content of this presentation useful today. Please do look at some of those extra resources I've pointed you towards, and if you've got any outstanding questions, please contact me on my e-mail address there.

NPIC: Northern Pathology Imaging Cooperative

Hello I'm Darren Trainer, a pathologist from Leeds in the UK and I'm here to present the Northern Pathology Imaging Cooperative. These are my disclaimers and our funding statement for the Northern Pathology Imaging Cooperative. 

The Northern Pathology Imaging Cooperative is a UK project to deploy digital pathology across a large geographic area and to use this infrastructure to develop and evaluate artificial intelligence in the real world.