Part of a Larger Whole

Part of a Larger Whole

Digital pathology can provide great return on investment – but only if laboratories are ready for a sea change

Digital pathology is expanding in daily practice. More and more groups are moving to a totally digital histopathology service, spurred on by the practical experience of the pioneers. It’s limited experience, of course – only a few years’ worth – but it has demonstrated that digital pathology can be a reality. And the fact that some brave people are working fully digitally now is key, because they’re encouraging others to take the same step. There’s a distinct change in attitude – nowadays, pathologists can see the speed and accuracy of digital diagnosis, and they recognize that the technology isn’t as distant or as scary as they thought. This is especially true in institutions where telepathology is necessary – for instance, if they need the services of a particular specialist, but don’t have one on staff. Overall, I’d say that, in the last three years, digital pathology has changed from a sporadic phenomenon to a widely known and increasingly popular option.

Watching your workflow

When I ask why laboratories choose not to adopt digital technologies, I often hear, “Well, we have other priorities.” They feel that they have other problems to solve – things like buying needed equipment or moving into new techniques. But in hospitals where the basic problems are already solved, digital pathology is a real priority. They’ve realized that it can help them optimize the way they work.

Unfortunately, the financial aspect rules much of the decision-making. Digital pathology offers a return on investment – but only if you’re ready to do it from a global perspective. You have to be willing to change your workflow and move people from tasks that are no longer needed (like delivering glass slides, or searching through archives) to ones that still are. Of course, it doesn’t all happen at once – you start with duplicate processes, glass and digital, and you have to make allowances for both. It can be difficult to convince those who control the purse strings to invest in digital storage without eliminating the physical. Nowadays, though, the options for inexpensive long-term storage are multiplying, so the main investment is in short-term storage – which requires much less capacity. With a sensible plan for storage, any laboratory can make a good return on digital pathology investment.

Why isn’t everyone seeing such a return? I often see laboratories focusing on a single solution rather than their global workflow. For instance, they’ll decide that they want a scanner for digital image analysis – but if it isn’t part of a bigger picture, it will be a very expensive solution to a single problem. You have to be willing to make a wholesale change; for instance, it’s a good time to apply Lean technology or Six Sigma. It’s a good moment to improve every single step of your laboratory’s process. It sounds like a lot of work – and it is a lot of work – but after you work hard for a few months, you get the payoff for years.

Serendipitous solutions

In my laboratory, we call our transition “serendipity” because everything aligned just right to make it happen. The most important thing, though, was our engineers’ willingness to take on a new project. We told them, “There’s one really complicated thing we’d like to do: manage all of the large images we capture using digital pathology techniques.” They agreed that it was a difficult task – and that’s why they wanted to work on it. We were lucky that they wanted to collaborate with us on it! It’s not enough to have a supportive organization or an innovative team that wants to transition to digital; you need good support from your computing department. Without them, we’d have been lost.

Ultimately, I hope we’ll end up with universal technology solutions. Things are a bit fragmented right now; each digital pathology company has its own formats and standards. When those companies realize that they need to use a common format, so that we can all view and manipulate slides regardless of how they were produced, things will become much easier. I understand that each company wants a proprietary solution, but I think it’s vital for them to understand that we – the pathologists – really need a single standard, like DICOM. Having to change from one format to another adds an extra step to the process, complicating it and potentially impacting patient care, and the only way around that is to share a universal format.

Once we’ve accomplished that, the next step is to address storage. Pathologists are being treated like we need a specific solution – but we don’t. We simply need the same kinds of solutions as radiologists, endoscopists and others already have. The problem is that there’s always more information to be managed. When the price of storage decreases, we allow ourselves to image in 10 planes instead of just one. Then we create 3D images instead of 2D. Then we increase the resolution. And with each change, the image size increases – and with it, the demand for storage. In the end, it’s up to us to be aware of what is useful for diagnosis. Do we really need 3D for everything? Do we really need to capture all our images at 60X resolution? Technology will continue to improve, but we have to find a balance between what we can do and what is worth doing.

A marriage of man and machine

With AIDPATH – the Academia and Industry Collaboration for Digital Pathology – we’ve been working hard to compare different image analysis solutions with each other and with manual methods. Although there are differences between manufacturers, we’ve realized that some of the discrepancies are due to human error. The pathologist selects what needs to be evaluated in each image, which means that no matter how well the technology itself works, there’s still a human factor involved that we can’t control very well. For instance, what if the user selects an area of in situ carcinoma instead of infiltrating carcinoma? or misses a region of a tumor image with high expression of a protein of interest? The conclusion I’ve reached is that we need solutions to help pathologists locate the most interesting parts of biopsies. For instance, it would be very interesting to design algorithms that can help inexperienced pathologists locate infiltrating carcinoma, or detect areas with the highest biomarker expression. Some of these types of algorithms are already in development, and they’re working very nicely so far. Perhaps this is the beginning of a beautiful partnership between humans
and computers!

Marcial García Rojo is the principal investigator in the EURO-telepath EU project, a principal researcher with AIDPATH, and head of pathology at the University General Hospital of Jerez de la Frontera, Spain.

Read more:

The Doubting Dollar
By Luke Perkocha

A Roadmap to the Future
By Alexi Baidoshvili

Techniques in Transition
Michael Schubert interviews David Snead

A Business Case for Common Sense
By Liron Pantanowitz