Aleksandra Zuraw
In the emerging field of digital pathology, Aleksandra Zuraw, DVM, Ph.D., DACVP, is one of the most prominent voices. Her journey to this position, however, was far from straightforward. When Zuraw was 12, she dreamed of becoming a veterinarian, picturing herself following in the footsteps of the British veterinary surgeon James Herriot, tending to “All Creatures Great and Small.” But a particularly frigid winter spent caring for calves convinced her to bring her talents indoors. What followed was a winding career path that would eventually lead her to the leading edge of digital pathology, a field with important ramifications for drug development given its essential role in accelerating and streamlining clinical trials and focus on disease pathology.
Zuraw’s journey reflects a broader shift taking place within the life sciences. While she once focused on treating individual animals, her work increasingly has a data slant. As an image analysis pathologist at Definies and later veterinary pathologist at Charles River Laboratories, Zuraw has been deeply involved in designing image analysis algorithms for clinical trial and preclinical study evaluation, working alongside computer scientists and pathologists at various pharmaceutical firms.
Getting the word out about digital pathology
In addition to her work as a practicing veterinary pathologist, Zuraw is the founder of Digital Pathology Place with backing from prominent players in the field such as PathPresenter, Pramana, Bionovation Image Cytometry, Proscia, Hamamatsu, Smart in Media, Aiforia, and Epredia. She is also the host of the Digital Pathology Podcast and author of “Digital Pathology 101: All You Need to Know to Start and Continue Your Digital Pathology Journey.” This diverse support from industry leaders underscores the recognition of Zuraw’s contributions to advancing digital pathology through education and outreach.
Although she had prior exposure to digital pathology, the pandemic accelerated adoption. “It wasn’t just nice to have anymore,” she said. “It was a must-have.” While pathologists traditionally worked in labs handling physical slides, COVID-19 restrictions complicated that model. “Suddenly, pathologists couldn’t travel,” she recounted.
Peer review, a cornerstone of many preclinical studies, became a logistical challenge. “Before I would be at the site, looking at the slides under the microscope,” she said. “The peer reviewer would travel to the site and spend one to three days to check if they agree with my evaluation results and report.”
The pandemic’s impact on digital pathology
Aleksandra Zuraw
In addition, shipping slides across borders during the pandemic was another challenge. “So the whole industry ran into timeline problems,” she said. “The only solution was to digitize those slides and let people view them digitally.”
“The pandemic triggered a lot of changes in the whole pharma industry,” Zuraw said. A growing number of peer reviews once done in person were conducted digitally. “This triggered a massive validation effort of those [digital pathology] systems for GLP (Good Laboratory Practice),” she noted.
In the intervening years, the adoption of digital pathology has continued to expand, evolving from an emergency measure to a more efficient way to work. For instance, its impact on preclinical research and development (R&D) has continued to expand while helping sift through the high volumes of preclinical R&D data, which can often eclipse that of other phases of drug discovery and development.
“Digital pathology is enabling preclinical pathology and R&D teams to do their jobs better,” Zuraw noted last year. “By shifting from microscope to whole slide image, scientists can work more productively and deliver more consistent results.”
Goodbye, boxes — Hello, digital efficiency
In the meantime, Zuraw’s workflows have become less cumbersome. “Let’s say I have a study with 1,000 to 2,000 slides, and I get two to four weeks to read them,” she said. “As a remote pathologist, I’d get huge cardboard boxes full of plastic boxes with slides. The biggest efficiency gain for toxicologic pathologists is not having to deal with those boxes.”
For pathologists accustomed to working digitally, the transition has driven significant changes in day-to-day operations. “My microscope is now more for my educational content online than for evaluation because I have everything shared with me digitally,” Zuraw said. And when she visits her native Poland, Zuraw has a validated workstation. “So I can work from there. It is fantastic,” she said.
Integrating AI and image analysis
The rapid embrace of digital pathology opened doors for integrating artificial intelligence (AI) and image analysis to further enhance efficiency and accuracy. While acknowledging that this technology is still evolving, Zuraw is optimistic about its potential: “We have identified the most common histopathological changes in different studies and started developing image analysis solutions for that,” she explained. These technologies rely on supervised deep learning models trained on labeled datasets to identify and quantify specific features in tissue samples.
The Digital Pathology Association has noted that the adoption of digital pathology has seen strong growth between 2021 and 2023, with reported growth rates of 37.3%, 31.4%, and 27.3% over the past three consecutive years.
According to a recent survey from Proscia (conducted by Atheneum), 82% of organizations that use digital pathology have implemented AI to date, and all of those yet to implement AI plan to do so this year. Among the AI adopters, 87% are using image analysis and two-thirds, 66%, are using process-automation-based applications.
Enhancing pathologists’ capabilities
“There are several algorithms for different organs that use supervised deep learning models,” she said. The benefits are already apparent: “It makes me faster and more consistent for those changes that we have algorithms for.”
Zuraw emphasizes that the goal is not to replace pathologists — who like radiologists are in short supply in many parts of the world — but to augment their capabilities: “Basically, it’s about working [reading slides] faster with the same quality and open the doors to additional insights from the digital slides.”
The subtleties of pathology work can cut down on tedium. “These are very subtle changes, like single-cell changes,” Zuraw points out. “So you don’t want to go too fast because you’re going to miss it.” Yet, speed remains a priority: “But you don’t want to be slow either, especially working for a CRO, because there are timelines associated with this — FDA filings and everybody wants to get their drugs developed fast.”
Paving the road to the future of digital pathology
While the digital transformation of pathology remains a work in progress, digital is “part of pathology now,” Zuraw said. “Maybe people who only need five years to retirement don’t have to bother, but everybody else needs to because that’s our world now.”
In the coming years, familiarity with digital pathology workflows could be like white-collar workers being adept with spreadsheets, word processing, and checking email — fundamental skills. “At some point, it’s going to be like, ‘Okay, you don’t know digital pathology? We’re going to find somebody else,’” she said.
Zuraw is quick to point out that, despite the short-term learning curve, digital pathology presents an opportunity for pathologists to influence. “If you can use the tools, then you have leverage,” she explains.
That doesn’t mean that adopting digital pathology workflows is always a linear journey. “If you embark on this journey right now, you can be sure that there’s going to be a lot of troubleshooting, a lot of trailblazing in your lab, in your team to manage the change, to make sure that everything is interoperable,” she said. Being willing to step outside of one’s comfort zone while finding new ways to work with vendors and users, purchasers, the image analysis team, the validation team become key. “It’s super interdisciplinary,” she said.
Filed Under: Data science, Drug Discovery and Development, machine learning and AI
