News Article

The future of pathology is digital

June 18, 2024
Colon tissue stained using the BOND RX and imaged with an Aperio digital pathology scanner.
Colon tissue stained using the BOND RX and imaged with an Aperio digital pathology scanner.

We spoke with Rob Monroe, Vice President and Chief Scientific Officer of Oncology, Danaher Diagnostics Platform, about how cutting-edge technology and AI-driven tools are transforming the art and science of disease diagnosis. 

The digitization of radiology was a huge moment for diagnostics. Why hasn’t digitization of pathology seen faster adoption and similar rapid uptake? 

The ability to digitally capture and share radiographs transformed the field. It removed the requirement that a radiologist must sit in the hospital, which meant, for example, that a radiologist could be sitting as far away as Australia reading cases during normal working hours processed in United States emergency rooms in the middle of the night. 

There are a few key differences between radiology and pathology that make the equivalent jump more difficult. 

One is that in radiology, you can jump straight from capturing an image to sharing a digitized version of that image, like using a digital camera instead of a film camera. In pathology, we still have to contend with the physical nature of the work: processing the tissue, producing a glass slide, staining the glass slide, and then, finally, digitizing it. There’s a complex workflow that must happen in the laboratory—producing a slide that must be completed regardless of whether the slide goes on to be digitized. As such, digitization actually adds to the complexity rather than reducing the number of steps prior to the pathologist’s review of the slides. 

Another difference is in the images themselves. Pathology images are ten to 100 times larger than digitized radiology images, so manipulating, sending and storing them all pose significant technological challenges. 

What’s changing to make digital pathology more feasible now? 

Simply put, the technology is catching up. Our capacity to scan, store, view and share high-quality images has skyrocketed over the last few years, and with AI and machine learning applications for pathology growing in parallel, the incentive to put these technologies into practice is only growing. 

We’re at an exciting inflection point where we have the tools and are beginning to see how much value they bring to pathology labs in driving more efficient workflows and improving diagnosis. Right now, we’re in the early adoption phase, hovering around 5-10%. We think it’s very possible to drive adoption to 90% over the next few years. 

Now it’s a question of how to demonstrate the technology’s value to labs and pathologists without significantly increasing their costs to drive broader implementation.

What’s the biggest practical obstacle to digitizing pathology? 

Two major things will need to fall into place to drive wider adoption of digital pathology. 

The first is reimbursement. Right now, getting payers to pay for new diagnostic technologies like digital pathology is challenging, which means early adopters may not see a return on their investments for many years because the evidence supporting reimbursement is still being generated. This is a front-and-center challenge for all of us working at the leading edge of this field: ensuring payers have the relevant data they need to make payment decisions that accommodate new technologies. 

The second is increasing acceptance among pathologists and labs themselves. In labs that are already working at maximum capacity, carving out time and resources to change workflows and adopt new technologies is an uphill battle. We need to demonstrate how these new tools will benefit pathologists in real world clinical settings, saving time and making better diagnosis both easier and more accurate. 

Every lab has a distinct set of circumstances that impact their decision to go digital. As more labs start to implement digital pathology, we’re getting helpful information about best practices. While a one-time, universal transition across the organization may sound efficient, some organizations take on a limited roll-out in one area or subspecialty (for example, prostate biopsies) to allow for a trial period of learning and refinement. This can be a successful model to demonstrate the power of the technology and drive buy-in among stakeholders: starting with a narrower scope and expanding once the processes are refined. 

How do we go about solving the reimbursement problem? 

We need to demonstrate the real clinical value that digital pathology can offer. We know it’s there—for example, the application of AI to improve the accuracy and reproducibility of immunohistochemistry scoring—so the need is more in collecting data and presenting it in a way that makes that value clear for payers. We’re deep in the data-gathering phase right now, and as more AI tools are incorporated into digital pathology workflows, the value is only set to grow. 

How do you see AI fitting into the pathologist’s workflow in the future? 

Digitizing pathology has tremendous value, enabling things like remote case review and consultation, case sharing and archiving, and more robust education. Layering on AI gives us the capacity to analyze complex images faster and more effectively than ever before. We can detect tumors and tumor subtypes, conduct quantitative biomarker analysis, characterize novel morphological structures and localize rare events. In short, it’s a game-changer. 

Our vision for AI’s role is as a kind of support tool. Much like with self-driving cars, the first role of AI will be more of an assistant. Someday we may be able to use the tools to augment or replace certain functions, but in the near term, the greatest role AI can play is as a partner to the pathologist. 

This could be as an initial data analyst, or in helping with all the parts of the pathology workflow that aren’t necessarily directly related to diagnosis, so the pathologist can focus on the most important work. 

Looking ahead, at Danaher, we’re actively pursuing collaborations with key opinion leaders (KOLs) in the AI world to further our involvement and ambition in this area. One exciting example is in morphology imaging that includes both digital pathology from Leica Biosystems and digital hematology from Beckman Coulter Diagnostics, where we are working with KOLs to predict genomic changes and therapy response of tumors through assessment of morphologic features. At Danaher, we refer to these areas where we can layer technologies on top of each other to create a sum greater than their parts as “group catalyzed value,” and we believe they offer an exciting glimpse into what the future holds across life science and medicine, including diagnostics. 

Why is it so important to digitize pathology?

Pathology is primed for a revolution. Tissue samples remain a valuable tool to diagnose disease, so we need to figure out how to speed the adoption of diagnostic methodologies including digital pathology and AI to these traditional tissue-based methods to increase their impact and improve patient care.  

From a practical standpoint, digitization could also dramatically broaden the workforce.

We’re facing a significant projected shortage of pathologists in the future, so tools to review images from remote locations will become even more important in recruiting and keeping talented providers. 

Digitizing pathology is a key step to ensure that we deliver the best possible information to providers and the best care to patients, with fast and effective diagnostics that lay a foundation for better treatments.