CEO SUMMARY: Now that the FDA has cleared two digital pathology systems for use in primary diagnosis, a growing number of pathology groups are taking up the question of whether and when they should adopt and use a digital pathology system and whole slide imaging. One pathologist with hands-on experience working with different digital pathology systems says that the technology is improving. He offers insights about the current state of the digital pathology market.
WHOLE SLIDE IMAGING SYSTEMS REACHED A MILESTONE OF SORTS in May when the U.S. Food and Drug Administration cleared Leica Biosystems to market its Aperio AT2 DX System for clinical diagnosis.
Leica’s product is the second whole slide imaging system to receive FDA clearance in the United States. In April 2017, the FDA cleared the Philips IntelliSite Pathology Solution (PIPS) for marketing. The PIPS allows pathologists to review and interpret slides prepared from biopsied tissue. (See TDRs, July 22, 2019, and April 24, 2017.)
But regulatory clearance of these two systems for use in primary diagnosis may not mean that whole slide imaging systems will soon supplant the glass slides currently used by the nation’s anatomic pathology labs. At their current state of development, use of a digital pathology system and whole slide images (WSIs) in a pathologist’s daily workflow delivers certain benefits, along with some offsets.
One pathologist considers that the current state of technology in digital pathology and WSI is improving steadily. “My sense is that the efficiency factor of about 20% longer to read a WSI best represents the current reality that WSI is in its early phases of adoption,” commented Richard Feddersen, MD, Medical Director for Immunohistochemistry and Anatomic Pathology for TriCore Reference Laboratories in Albuquerque, N.M. “Digital pathology vendors have and will mount reasonable arguments as to how this will improve over time.
“As an example, a standard two button-roller ball mouse (as used in these Leica studies) may not be the ideal peripheral for navigating the digital slides,” he explained. “I’ve seen efforts to substitute a roller ball or touch screen. Both would be potential enhancements for at least some diagnosticians. One idea I haven’t seen tried is to adapt a high-end joystick-type game controller to the software, creating an environment akin to a flight simulator.”
Pathologists interested in comparing the workflow of digital pathology systems against viewing slides through a microscope will find many advantages and disadvantages to each method, Feddersen explained. He was his site’s principal investigator for the study Leica Biosystems did to support its application for FDA clearance.
In his evaluation of Leica’s AT2 DX System, Feddersen found the Leica system compared favorably. Not only did he serve as the principal investigator for the AT2 DX, but previously he was one of the reading pathologists for another WSI system that, under the terms of his agreement with the second vendor, he could not name.
Assessing Image Quality
“The image quality is at least comparable to the other systems I’ve seen,” he said of the AT2 DX. “Generating a quality digital image from material that is as optically complex as a microscope slide at a variety of magnifications is a huge technical achievement.
“And with all these vendors, you can almost take for granted that you’ll get a quality image at either 20X or 40X magnification,” he added. “They all seem to perform comparably. Put another way, image quality is a basic prerequisite for entering this market.”
From the Leica study, Feddersen provided four key takeaways. The first, as he mentioned, was that the image quality was comparable to that of other systems.
“The second takeaway is that the Leica hardware works,” he said. “By that I mean there was little need to rescan slides once they went through the instrument the first time. In the past we’ve reviewed competing systems where this wasn’t the case, although—to be fair—I’m not up to date with all of them. Suffice to say that a system would need a slide rescan rate of under 1% as a requirement for persistence in this market.
Throughput Rate for Scans
“Here’s another detail that I’d add: When assessing a scanning system’s potential throughput rate from a stained slide to a digital image,” commented Feddersen, “consider whether the glass slides need to be reracked individually by hand—which is far less preferable to a scanner that accepts slide racks compatible with your automated stainer.
“A third takeaway would involve my attempt to encapsulate all the wonderful possibilities that become available when you gather and archive a whole slide digital image,” Feddersen commented. “I’ll begin with image analysis: both the Leica and Philips systems offer robust image-analysis software that layer right into the digital image product and can interface with the major anatomic pathology LIS systems. To me, they both appear to be quite satisfactory, at least in demonstration mode.
“Beyond image analysis, a competitive software package enables any number of colleagues to conference on a case from remote sites or mitigate the cumbersome aspects of an extra-mural consultation,” he said.
Benefits of Using Digital Path
“Software also allows pathologists to prepare well-organized interdisciplinary presentations in less time with more content, to prepare teaching slide sets much more readily than before, and to quickly gather and permanently associate important measurements and other annotations with a slide, thus documenting one’s reasoning,” he added.
“For labs with pathologists working at scattered satellite locations, software can solve the logistical headaches of delivering materials in a timely way, something that is often a big problem in the afternoons,” he said.
Pathologist Offers Candid Views on How Firms Could Boost Performance of Digital Pathology
BASED ON HIS EXPERIENCE evaluating and working with digital pathology systems, pathologist Richard Feddersen, MD, has a wish list for what these vendors should address next. As the Medical Director for Immunohistochemistry and Histology Co-Director for TriCore Reference Laboratories, Feddersen suggested vendors make enhancements to slide viewers, image analysis software, file storage, and artificial intelligence.
Slide viewers. “Digital pathology system vendors need to keep working on their slide viewers, the interface that enables a user to display and navigate the histologic image,” he said. “The current default is to use a standard mouse, but there may be more intuitive and efficient alternatives, and a single solution may not be best for all diagnosticians.
Image analysis. “As it relates to quantitative immunohistochemistry, image analysis is already available at a high level,” he commented. “What remains is to continue fine tuning algorithms to deal with problem cases: such as poorly cohesive tumors, tumors with heavy inflammatory infiltrates, and suboptimal histologic preparations, to name a few.
“Image analysis may have unrealized potential for discovering very rare events, example, or single metastatic cells in lymph nodes,” he commented.
Throughput. “The sequential leaps in slide scanning automation and throughput over the past 10 years have been remarkable,” he said, adding that vendors should, “Keep it up!
“Also, a low-tech yet substantial throughput enhancement has been the standard slide rack used between two of the robotic H&E stainers on the one hand, and the Philips and Leica scanners on the other, eliminating the need for slides to be individually unloaded and reloaded by hand,” he explained.
“It would be great if there were a standard rack across the entire histology robotics industry,” he said. “The onus for this feature rests mainly with the designers of the automated stainers.
File storage. “The vendors must develop and constantly refine a menu of image file storage solutions which cost-effectively meet the needs of individual clients,” he suggested.
Artificial intelligence. “While artificial intelligence may or may not ever be able to replace a trained human microscopist, it would certainly be interesting to watch an advanced AI system’s differential diagnosis appear along the side of the screen while examining a digital slide!” said Fedderson.
Value in Digital Archives
“And finally, depending on a lab’s case volume and scanning capacity, there is the potential for a transition to digital slide archiving, eliminating once and for all the labor-intensive physical storage and retrieval of glass slides, problems that we’ve always known and not loved,” Feddersen commented.
“I should emphasize that while there is only a small handful of big firms marketing high-throughput slide scanners, there are at least two handfuls of small software enterprises innovating ways to view standard digital image file formats, and perform all the image analysis and other functions I mentioned above,” he added.
“Some of these companies are getting into diagnostic artificial intelligence applications,” he said. “Competition in this field appears rather intense right now.
“My fourth and last takeaway is more cautionary,” he suggested. “A quality whole slide image is a densely granular thing, and the data storage and retrieval requirements can be huge, especially if the goal is 100% digital slide archiving.
“Just doing a thumbnail calculation, I’d say that to scan and archive all the slides our lab would need 150 terabytes of new file storage each year,” he estimated. “Therefore, pathologists have a lot of calculations to work out in advance, such as the mix of onsite and cloud data storage and expectations about retrieval time for current, recently archived, and older cases.
“From what I can tell, the WSI vendors could be a little more proactive in offering prepackaged solutions modeled to fit client needs,” he commented. “If they don’t do so, some of the small software startups will. At present the plan seems to be an assumption that ‘Our IT people will work with your IT people,’ a phrase that triggers alarm bells for many of us in laboratory medicine.”
A Question of Throughput
After describing his four primary takeaways, Feddersen addressed the key question many pathologists have: How does the throughput of the AT2 DX perform in a busy practice?
“I would say the AT2 DX is comparable or superior to other workflows, although the current Philips system is doing so at a higher resolution,” he commented. “That said, Leica and Philips may be leapfrogging right now, and I don’t have enough experience with Hamamatsu and other vendors to comment on them.
“There are several practice scenarios where one or more of these instruments should fulfill a group’s long-term needs,” he added. “For example, any pathology group would benefit if all the doctors are ready to transition to routine digital diagnosis at once. Or, failing that, a group would benefit if they all work in the same office complex or if digital slide archiving is deferred to the future, and only some subset of the slide output is to be scanned. These instruments also benefit a lab that is able to finish all or most of its slide output on the night shift.”
A Paced Adoption of DP
TriCore’s 20 pathologists in Albuquerque are in two practices spread out over a large metro area.
“They want to adapt to the new digital format at their own pace, which I personally think is a good idea,” Feddersen commented. “They view 1,200 to 1,500 routine slides per day, and then generate around 500 special requests.
“We never have all this work done by 8 am,” he added. “We would love to transition to digital archiving from glass-slide storage. We’d prefer to scan all the freshly-prepared slides at the core lab before delivering them to the satellites, although gathering them up from the periphery after diagnosis is a less elegant possibility.
“So, we have a long—and some might say an unreasonable—wish list, for which the throughput capacity of today’s current digital pathology scanners would be hard-pressed to fulfill,” he concluded. “I don’t know how many anatomic pathology operations around North America share our needs, but I suspect there may be a few.”
Because of his hands-on experience with multiple digital pathology systems and scanners as part of formal studies and evaluations of this equipment, Feddersen’s observations and recommendations should be useful for any anatomic pathology laboratory considering when and how to take the plunge and “go digital” with its daily workflow.
Contact Eric Carbonneau at 505-938-8470 or firstname.lastname@example.org.