Teleradiology and AI — An Interview with Medica, UK’s leading teleradiology firm

Qure AI
13 min readJul 12, 2021


In this interview, Healthtech firm’s Reshma Suresh speaks with Samantha Davey (SD), Head of Program Delivery at Medica Group Plc, and David Evans (DE), Functional Architect at Medica Group Plc to learn about their experiences working with various AI projects in Medica. Medica has been one of the leading teleradiology providers in the UK and has been supporting clinical radiologists since 2004 to deliver the best patient care. Medica’s service is in use by over 100 NHS hospitals.

RS: Working with Samantha & David has been a great experience, I would like to thank both of you for joining us here, let’s start by sharing this learning with our audience.

DE: Thanks for inviting us to the podcast here. I’ve been with Medica since 2008, currently, I am working as a functional architect. I spend most of my time working on the Future Tech program, it is a series of projects that we have in place that is going to fundamentally underpin medical strategy for future growth, and the deployment of AI from has been part of that program. Before working with Medica, I was a medical physicist in a diagnostic radiology department.

SD: Thanks David and thank you Reshma for inviting us today. I am the head of program delivery at Medica. My background is in project and program management. I’m one of 1000 chartered project professionals globally and working on something interesting and innovative like the adoption of augmented intelligence into our workflow has been a privilege for me.

RS: Thank you very much. Medica is one of the largest teleradiology companies in the world. Can you tell me a bit about Medica? What has been Medica’s offering and what makes Medica unique?

DE: When I started with Medica in 2008, it was still quite a small organization based out of a town called Battle, South of England. In 2014 we shifted our headquarters to Hastings. In 2017, we became a listed business on the London Stock Exchange.

We provide many different types of services — one of them is an elective service where clients can send the images for us to report with SLAs that range between 24–48 hours.

We also have an emergency reporting service called NightHawk, which is available 24/7 and reports back in about 23 minutes. Within the Nighthawk, we have some prioritized services, like stroke pathways and major trauma.

We also provide some more specialist services such as reporting of cardiac CT, PET CT, and DEXA reporting. We’ve now grown to have around 500 plus reporters based in the UK Australia, New Zealand, and in places around Europe. We have 100 plus clients now split between the NHS and the independent sector. One of our recent purchases was an Irish company now called Medica Ireland and a clinical trials company in the US called RADMD.

We see ourselves as a strategic partner to our clients. We’re there to supplement the clinical radiologists’ own internal resource, their own internal skills.

One of the unique aspects of Medica is the fact that we are clinically led. Teleradiology is heavily dependent on technology, but we do see ourselves being a clinically-led organization that’s underpinned by technology rather than a group of people who focus more on the technological aspect. We made sure that our reporters have access to all the clinical information that they will need to provide the radiology report. You can see all the previous history in terms of the reports and the imaging as well, and I think that’s unique within the teleradiology domain.

RS: Medica is it’s a large organization with 500–600 reporters who are specialists reporting the scans. Apart from that, you have other supporting staff working with the NightHawk portal. At what point, and why did you consider adopting qER in the workflow?

SD: We’d been considering the potential and possibilities around using AI for some time. We then began to have detailed conversations with several organizations and Qure was one of them. We began our partnership with Qure and began to work on the adoption of the qER algorithm in earnest in late spring last year. We chose to work with qER because, within the acute workflow, intracranial hemorrhage is one of the key things that we need to be aware of when it comes to prioritization of those studies, and as David said, our turnaround is incredibly fast from the point of receipt of the image to the report being returned, it’s less than 25 minutes. So, a tool that can speed up that prioritization was very important to us.

RS: Getting timely treatment, access to diagnostics to the patients is the vision that every teleradiology solution would try to solve or try to achieve. What are some of the key pointers a customer client hospital would consider before going with a teleradiology provider like Medica?

SD: We believe that the teleradiology partner must be clinically led. They should have a good level of clinical governance. They should see themselves as a clinical company underpinned by technology, not a technology company that happens to be dabbling in health care. You couldn’t have teleradiology without the technology but what we’re here to do is to provide solid reporting to ensure that the patients get the right care at the right time in the right way. That needs to be the primary factor in terms of choosing the right teleradiology partner.

DE: Choosing the right teleradiology company is vital. Clinical quality is important as you should consider them from the clinical perspective rather than thinking about the technological tool. Some factors to be questioned are how does the teleradiology company select the reporters that they are going to use to deliver your work? and how do they measure the quality? and how do they maintain and ideally improve the quality over time?. For lots of clients, an important consideration is the breadth of services that are available from the teleradiology provider. You need to ensure that you’re going to get consistent delivery against your needs, and you can only do that if the teleradiology company does have the breadth of services and reporters. Reputation is vitally important in teleradiology. The radiology community talks within itself, so I think it’s important that you maintain that high reputation. Potential customers should be looking at how quickly and efficiently the partner can provide the report.

RS: Can you tell us a little bit about NightHawk and the challenges that you have faced while operationalizing the NightHawk portal?

DE: NightHawk is a service we provide to our clients to provide emergency reporting. NightHawk provides out-of-normal hours service with fast turnaround times. We’re currently providing that kind of emergency reporting for up to around 50 hospitals per night and at peak times we have around 25 radiologists who are not just there to report on the actual CT scans that come through but also to discuss the cases with the doctors that are making use of the NightHawk service. A reasonable percentage of cases will be where our radiologists will advise that emergency imaging isn’t required. Our radiologists will then provide a written report back into the client information system. As soon as the reporters provide and authorize that report, it’s available for the client to access straight away. The speed of delivery of that report is vitally important when you think that these referrers could be managing patients for major stroke trauma. If ever one of our radiology experts providing the service finds something that’s urgent or unexpected and they need to notify the referrers of that for some urgent treatment, then they’ll call the referrals back at the client end to make them aware of the case and to talk about the urgent factors or the urgent clinical need.

The activity levels for NightHawk vary but are typically between 600 to 1000 patients per day, and the vast majority of those come during the NightHawk hours rather than during daytime hours. Working for Nighthawk we will be receiving everything from CT, head CT, brains to full-body CT or head, neck, chest, abdomen and pelvis, major trauma scans, et cetera. And some of these scans have got very large volumes of data associated with them. When you’re working these over wide area networks, then it takes longer to transmit the image data. The longer it’ll take us to receive it and provide that report back to the clients, so we do work very closely with the clients to try to ensure that they are only sending the data that we need and not wasting bandwidth or transmitting data that we don’t need or not going to contribute to the radiologist report.

SD: David can you share some insights on how we’ve integrated qER into the Nighthawk portal because that’s been a specific piece of work that the software development team has done, and obviously that was fundamental to enabling us to get qER up and running.

DE: The software development team have built the NightHawk portal and everyone within the NightHawk process, from our clients who are notifying us of the need to use our service, through to our NightHawk Department who are managing the service and prioritizing studies and making them available to our radiologists and the radiologist themselves who are providing that radiologist output, they all can access this NightHawk portal. What we’ve done with qER is we’ve integrated the outputs from the qER into that portal so that if it does flag an ICH in the middle of the night, it’s available to the allocations team so they know that they need to prioritize that study and make it available to one of our reporters just as quickly we can. Secondly, it’s visible also to the reporters within the NightHawk portal. The reporters can see that the CT head scan has been through the qER algorithm and identify a potential ICH. We worked with you guys at Qure to implement an additional algorithm that will suggest if there’s an infarct in the study and again, we’ve integrated the output of that into the NightHawk portal. I guess it’s akin to how you might integrate the qER algorithm with an NHS organization or a hospital organization where you would send HL 7 message through and that’ll be processed and included within the resulting output. It is good that you’ve got the flexibility to not just have that single HL 7 model.

You’re also able to provide adjacent messages, adjacent outputs as that’s what we use, and we’ve integrated that within our NightHawk portal. We were also able to integrate the outputs from the qER within our PACS workflow to get the images from PACS to make sure that the results from Qure get back to us and are put right in front of the people that need to see them as quickly as possible. A turnaround time of 23 minutes on average is quick and we make sure we can maximize the number of times that the qER output gets in front of our NightHawk departments and our radiologists. We are now fine-tuning the system with you guys for even quicker allocation in particular vital cases.

RS: What were your thoughts before we started doing this? And how do you think it panned out?

SD: We planned the project before we knew we were going to go into lockdown, so David and I spent quite a lot of time thinking through how we’re going to do the proof of concept, what our success criteria were, and how we would measure the success of the project, who we needed from a medical side to be involved in the project, what skills, capabilities, resources we needed. When we started working with Qure you were always responsive. Qure helped us with the additional algorithm to solve the issue of false positives due to infarcts. It was an absolute breath of fresh air to us to be working with a partner who genuinely entered a real spirit of partnership across the project. We are all doing this with a desire to improve our knowledge, to improve service and we’re operating with time differences. And we’ve been conscious of the fact that sometimes you have been working for hours but you’re still there on the platform, you’re still working with our infrastructure guys to try to resolve any of the problems that have come through and that’s awesome. The other interesting thing about implementing qER and the same for almost any other algorithm is that you can say the quality of the algorithm is good but mainly the challenges are around implementation and that is something we needed to be aware of beforehand.

DE: The main difficulty is not in getting your DICOM images off your PACS quickly enough to fit into your workflow or about the issues you might face in terms of taking HL7 or adjacent responses and making them available to the right people at the right time. I think it is something that we have benefited from your team’s responsiveness. One of the really nice things you can say is that it doesn’t feel like there’s a 4.5 hour time difference between us, and you guys and it’s been fantastic to work with you guys on that.

RS: When we look at qER and Medica, we see three points that we always consider. One is Medica as an organization. What does Medica want to achieve from qER? The second is the reporters and different specialists who are consuming these results, and the third is the clinics who are not directly interacting with Qure’s product or qER but they are also in a way affected in terms of improvement in turnaround time. Can you share some thoughts around what has changed before and after the introduction of qER from the 3 different perspectives if possible?

DE: qER has been a massively valuable tool for us and it’s because of the sheer volume of scans that we need to report for. For NightHawk during our busiest hours, we might receive four or five CT head scans in a couple of minutes. Before implementing qER we’d allocate those scans First in-First out method without any additional prioritization. If the client notified us of a stroke, then it’s on the stroke pathway and there would be some prioritization there. But essentially, most of the allocation is first in first out. Now after implementing the qER algorithm, it can evaluate those five scans in that period and let us know whether one or more of those is suspected as having ICH. If it does, we can then prioritize to allocate that scan for the next available radiologist. All the stakeholders that you mentioned are being benefited by qER. It benefits Medica because we are improving service to our clients knowing that we can allocate those studies with additional prioritization. The fact that we can prioritize the allocation of the scans where there is a suspected finding and the reporters can then prioritize the reporting of that scan, knowing that the algorithm has picked up on that suspected finding allows us to provide a better service to the clients and ultimately the patients.

RS: How is the software adopted and the confidence of your end-users in using that solution? Could you shed some light on some of the feedback that you have received from the end-users?

SD: This is one of the things that we’ve been mindful of and for us engaging with our reporter community is hugely important. We’ve got probably between 500 and 600 reporters now across the globe and our ability to provide the levels of service we wish to provide and hope to continue to provide to our clients is entirely driven by the quality of our reporters. We collected data with every single study and also pieces of feedback from reporters as part of our formal evaluation of the algorithm. We’ve had very encouraging feedback, 81% of those reporters say that they agree with the findings, and they’ve trusted in the algorithm. 71% of those who have used the algorithm said that they’ve found it helpful in terms of both prioritization and reporting.

I’m going to read some comments that we’ve got from our reporters:

“The algorithm is excellent”

“Highly likely to reduce the chances of missing a small amount of intracranial hemorrhage”

“Very good on the whole; comes on its own for the smaller bit bleeds that can be easily missed” and

“it’s very useful in identifying patients to get higher priority in reporting promptly.”

There’s has been a huge sense of recognition from our reporting community. The doctor’s clinical expertise determines what goes into the report but has that prioritization has been helpful for them.

RS: Based on your experience working with Qure or generally across other innovations, do you think teleradiology companies should bring in more innovation in their ecosystem?

DE: I think all companies should be looking to be innovative and teleradiology is no different in all settings. The fact that we met you guys to implement the qER algorithm and the underway Future Tech program, which is an exciting set of projects, are essentially examples that Medica has been innovative to make improvements. We try to innovate through process design as well as implementing new technology. And I think teleradiology companies should have the same mindset.

RS: I hope that whatever you have shared with us today will be helpful for organizations looking to adopt innovations. Thank you very much for that. Anything else that you want to share with us?

SD: I think the one thing I would urge anybody who wishes to look at this sort of project is to really think clearly about what benefits they wish to achieve from it. If the organization thinks that adopting an algorithm is simply plug and play then they’re going to be quite surprised, and they need to understand the wider scope of the project before beginning it.

DE: My advice would be to engage all the relevant people from within your organization and not just the end-users. We’ve had a project team for the AI project which includes infrastructure, software development, clinical governance, radiologist liaison team, service delivery, and other people from the PMO and Future Tech perspective. We pulled in people from across the organization as the impact of implementing an algorithm like this or another AI solution will impact so many parts of the organization as it impacts the entire process and the workflow. My final words would be to pull the right people at the beginning, help them design and build the solution and processes with you.

RS: Thanks a lot for spending your time and sharing such insights. We really look forward to working with you more closely and making this huge.



Qure AI is a breakthrough AI solution provider that is disrupting the radiology ‘status quo’ by enhancing imaging accuracy and improving health outcomes.