Journey to the Center of the Digitization Dilemma
Can two guys in a garage really shake-up current thinking on the transition to digital pathology?
Alec Hirst |
At a Glance
- Digital pathology is often approached as a minor adjustment to established practices, rather than as a substantive workflow change
- The major challenge in digital pathology at the moment is the storage and management of massive amounts of data
- To make a successful move to digital, it’s important to have pathologist buy-in, LIMS integration, and scalable, future-proof technology
- It’s not the solution to every problem, but digital pathology offers the chance to improve workflow efficiency with manageable cost and effort
In 2016, Alec Hirst and Chris Evagora, in partnership with their clinical leads Luis Beltran and Dorota Markiewicz, set up their company, Pathognomics. Their vision? To design and deliver a fully end-to-end digital histopathology laboratory to prove that whole digital pathology can work in a diagnostic environment – and that it can be implemented in a timely and cost-effective way. A lofty goal perhaps, given that Hirst and Evagora were working out of a garage rather than the shiny, fully outfitted building of an industry player…
Nevertheless, in under a year, they had built a wet laboratory, designed and implemented a diagnostic digital solution, successfully met Care Quality Commission (CQC) and National Health Service (NHS) Digital Information Governance (IG) Toolkit standards, and have been offered ISO15189:2012 accreditation. The laboratory has been operating diagnostically for six months and the team have completed 4,000 cases with two NHS trusts, as well as dentists and private healthcare providers. To learn more about their workflow and how others can learn from their approach to digital pathology, we sat down with Hirst.
What’s wrong with existing lab workflows?
The current approaches to laboratory practice are well-established, with some elements that have hardly changed at all over the last few decades. There is a drive to improve laboratory practice through automation tied to digital aspects, such as barcodes and macro-images, but these changes have been slow because, until recent years, laboratory quality assurance in the UK was captured under Clinical Pathology Accreditation. With the advent of international standards (ISO15189:2012), there has been improved assurance – but it has been more of a migration of established practices than a substantive change to laboratory assurance. As yet, no international standard has been developed to address the changes in digital aspects of pathology, particularly digital slide management. In addition, there is a lack of integration between regulating bodies – particularly with respect to data management and data security, which is assured under other legislation requiring skills outside of those needed in a traditional laboratory. This creates a need for more training, new staff members, or both, which can make it difficult to migrate a laboratory to a fully digital workflow.
To make such a migration work, we need to solve our key problem: data management. Digital pathology creates lots and lots of data. Where do you store it all? How do you retrieve it when you need it? Cloud-based solutions are both popular and important to delivering digital pathology, but the wrong type is expensive, and one cloud solution does not fit all. In trying to find the perfect solution, we have concluded that a lab should ideally use a mix of local hardware and cloud-based platforms to manage day-to-day operations, store digital slides, and backup information.
Another significant hurdle is the human factor. People often find it difficult to accept change, especially a transition as major as moving from traditional to digital pathology – and even if everyone is on board, there’s still a need for extensive training and education in the new workflow. To compound the difficulty, laboratory information management systems (LIMS) and image archiving and communication systems are not designed specifically for histopathology, meaning that they can be difficult to use in a tissue laboratory, and may mismanage data in a way that complicates the workflow or introduces the potential for error. Finally, it can be extremely difficult to obtain the proper accreditations for a fully digital workflow. Working with emerging technologies may mean that there is very little literature to draw on – trial and error can become your best friend.
How have you tackled those challenges?
We have designed and built our own LIMS to knit together established laboratory practice with digital tools. It’s designed specifically for histopathology and is a fully integrated digital platform validated and verified under our ISO15189:2012 schedule. In addition to diagnostic data, it manages staff training, ISO documents, reagent platforms, and finances, giving us a complete laboratory workflow that improves efficiency and significantly reduces our software overhead. In designing our LIMS, we have addressed all prerequisites for digital pathology under ISO15189:2012, IG Toolkit, General Data Protection Regulation (GDPR), and CQC:
- Quality control of images and patient reports,
- Responsive diagnostic environment,
- No local installation (minimizing user impact),
- Fast image transmission within IVD-CE controlled software technology, and
- Cost-effective image presentation and archiving in line with recommendations (currently 10 years for all images used for diagnoses).
Our goal was to present a seamless integrated digital product to the user – from laboratory management of training and education, ISO documents, and reagents to reporting and digital image management. And because the system worked so well for us, we now present it as a service to clients, removing the need for them to invest in technology and train their own workforce in the management and operation of digital histopathology solutions. Of course, larger laboratories may have the funding and resources available to handle their own transition – but for the many smaller labs and organizations who don’t, outsourcing is an efficient way to make the move to digital without overcommitting.
We have no on-site pathologists at our laboratory; all diagnoses are undertaken remotely. As you can imagine, without a digital solution this would require an unmanageable logistical nightmare of couriering slides to many locations! Not only would it be massively inefficient, but it would carry what we consider an unacceptable risk of loss or damage, not to mention a significant time penalty. The success of our model relies on our integrated LIMS to manage, track, and host a platform for digital reporting – so full integration between lab and LIMS is critical.
Our model is designed to be scalable, not just with workload, but also with emerging technologies. We are already in the advanced stages of integrating our LIMS with digital analysis tools. In a “big picture” sense, we have focused on capitalizing on the latest software and hardware to keep our costs to a minimum while future-proofing our technology. With the software we have implemented, we are well-positioned to benefit from future large-scale changes, which will be driven by data management, cloud-based platforms, and regulatory pressures on data and patient information.
Last, but not least, we have integrated our digital pathology workflow into our quality management system (including our surgical audits, digital validations, data management processes, and our obligations under IG Toolkit and the new GDPR), which ensures that our digital solution is quality-assured and accredited under ISO15189:2012. It was no small undertaking, but a valuable one – and one that we recommend to any laboratory moving to a digital setup.
Even with our current setup, we are by no means finished. In my opinion, we’ve now mastered digital pathology – so our eyes are firmly fixed on the next step forward: AI and deep learning. That kind of technology is much closer than one might think. We are paving the way internally and have been in discussions with AI and augmented pathology companies from the start to assist in deep learning and the implementation of AI apps as a tool to help our pathologists in their daily work. Exciting times ahead!
What did you learn along the way?
Beware of two guys in a garage! As a startup and small enterprise, we are a driven team – but, all too often, driven by budget. With only finite resources to call upon, we are forced to be novel in our approach to issue resolution. The good news: we can share those new and exciting ideas for everyone’s benefit. We are delivering a truly digital diagnostic laboratory that we hope is affordable and accessible to the NHS and to the wider community – just two guys in a garage!
By building our own platforms, we can be very accommodating to change. We have many routes for our clients and our team of pathologists to communicate with us, and we take all of their feedback onboard, modifying our systems to accommodate their needs. Inevitably, we run into technical and data protection elements, but our size allows us to make adjustments quickly. Our pathologists find our LIMS a great place to work and have said they consider it the best platform they have ever used.
It’s not all roses, though; we have also encountered some issues that I think are unique to micro-businesses like ours. The problem is a misalignment between the NHS drivers on engaging SMEs to supply services and innovations (something to which the NHS has expressed a commitment) and the benchmarks set by the NHS procurement processes. Pathognomics has failed two procurement opportunities at the pre-qualification questionnaire stage because benchmarks are too difficult for a micro-SME. It may not be intentional gatekeeping but, unfortunately, that is the outcome. But you can’t expect innovation from SMEs to drive change, savings, and improvements, if the benchmarks for procurement lock them out in favor of more traditional business structures and sizes.
Taken as a whole, the most surprising thing to me is that our solutions don’t already exist. In a way, though, our journey has given me insight into why: “digital pathology” doesn’t just mean the slide itself; it means the whole laboratory workflow. There is still a misconception in the industry that going digital will fix all the problems pathology currently faces. Our view, and one that seems to be gaining ground, is that digital pathology is a solution – but not the solution you may think it is. It requires a rethink of the traditional model, an acceptance of where the benefits lie, and an understanding that things need to change – something that takes time and skills that the sector currently lacks.
What’s your advice to others making the move?
Employ a project team comprised of all key stakeholders, from laboratory management and pathologists to IT and software experts. Moving to digital pathology is a complex change management project. We hear many stories where digital pathology is unsuccessfully implemented and, for the most part, it seems to be because of a lack of adequate project management. Expecting a laboratory manager or an enthusiastic pathologist to deliver a project of this magnitude will result in failure. They simply have too much work to do already – they don’t have the additional capacity to manage a huge project on top of their everyday tasks! They do, of course, need to be consulted and engaged, but the project needs to be managed independently of the day-to-day operation of an existing laboratory.
I would encourage everyone in the pathology sector not to lose sight of where the benefits of digital pathology lie – and to be aware that it is not as costly as perceived! Some mistakenly believe that digital pathology will fix all of our issues – notably, the shortage of pathologists. In my opinion, digital pathology doesn’t fix anything; rather, it is a tool that can be employed to increase the efficiency of the laboratory workflow. It helps, of course, that our pathologists can operate geographically independently of our laboratory; it means we always have someone available to ensure our turnaround times are met. In addition, we operate a model we call “fractional pathology,” making efficient use of our team by only purchasing the specialist pathologist for the work on demand. Not only is this cost-effective for our clients, but it also gives access to all specialist pathology as needed to meet demand – a different approach to a traditional locum model. Finally, data storage is a bigger issue than I think is currently understood – but it still doesn’t cost the thousands of pounds per terabyte offered by existing suppliers. It costs pounds and pennies, something we’ve proven in the course of our own work.
Digital pathology is not a panacea – but it’s a far more feasible and efficient option than many labs believe, and I encourage all of them to consider jumping on the bandwagon!
Alec Hirst is Program Director at Pathognomics, Huntingdon, UK.