Has Its Worth Been Proven Yet?

In a nutshell...

• Pathology and radiology aren’t equal, and the arguments for radiology’s digital transition don’t necessarily hold true for pathology’s
• Digital pathology holds promise for image analysis and quality assessment, but other applications don’t make financial sense – yet

In a nutshell...

• Pathology and radiology aren’t equal, and the arguments for radiology’s digital transition don’t necessarily hold true for pathology’s
• Digital pathology holds promise for image analysis and quality assessment, but other applications don’t make financial sense – yet

In a nutshell...

• Pathology and radiology aren’t equal, and the arguments for radiology’s digital transition don’t necessarily hold true for pathology’s
• Digital pathology holds promise for image analysis and quality assessment, but other applications don’t make financial sense – yet

Digital pathology is making headline after headline as increasing numbers of laboratories adopt new technologies. But is it really saving them money? Five experts provide their views on the business case for digital pathology, and whether or not they think the transition is a worthwhile investment at this point in its evolution.

The Doubting Dollar

With the introduction of whole slide imaging (WSI) over a decade ago, “digital pathology” burst onto the scene with much fanfare, yet has had relatively slow market uptake. Let’s find out why…

Eight years ago, I spoke about the business case for “digital pathology.” Here, I will revisit the topic with the hindsight of the last eight years. Where does the business case exist today?  What are the key factors to consider before adoption? And how should decision‐makers react to the ever-increasing “push” to go digital?

Let’s begin by clarifying the term “digital pathology,” because it can have broad or narrow meanings. I define it broadly as “the use, including display and analysis, of digitally acquired images of pathologic specimens (gross, histology or cytology) to accomplish a clinical or research objective.” By this definition, digital pathology can range from photomicrographs taken with a microscope‐mounted digital camera, to remote viewing of video images, to digital whole-slide imaging (WSI) – this encompasses viewing and interpretation of the digital images by a human being, as well as manipulation, analysis or interpretation of digital image data by software.

Being mindful of this broad definition is important, because many think only of WSI technology as “digital pathology,” which misses compelling business cases for other uses of digital images. However, WSI is the digital technology that requires the largest institutional commitment and investment, so it is WSI that is the focus of this discussion.

The radiology–pathology comparison story

To understand the current state of WSI, it is useful to consider digital radiology’s adoption story. When WSI was first introduced, one often heard, “Look how digital imaging transformed radiology. Soon, pathology will also be all-digital – it’s inevitable!” But drilling down on the transformation of radiology from film-based to digital, it becomes apparent that the comparison of WSI to digital radiology, as a transformational technology to move pathology from slide‐based to digital, doesn’t go far. The hurdles to WSI adoption by pathologists are similar, but the benefits are not comparable.

Digital radiology did four major things that advanced the business case for that technology

1. It ultimately removed costs from the system.

True, the new digital imaging technology required substantial startup and maintenance costs, including digital‐capable X‐ray and CT scanners, high‐end workstations and display screens, abundant digital storage, high‐bandwidth networks, and sophisticated software for processing, analysis and image storage and retrieval (PACS), as well as trained IT staff to install and maintain these systems. But the transition also eliminated many costs forever – including all of those associated with developing and printing film‐based studies, toxic chemical disposal, maintenance of film libraries, and all associated staff costs that went with film‐based radiology.

2. It improved the workflow of radiologists.

Digital technologies markedly increased radiologists’ efficiency and mitigated the impact of a then-looming shortage of radiologists. Pre-digital, radiologists sat in front of giant film alternators, large devices loaded with hundreds of films for scores of patients, which mechanically sorted and moved them into position for viewing. This was slow, inefficient and labor‐intensive, but still much faster than digging through piles of envelopes on a table or film library shelves. Once digitized, current and prior studies could be called up on a workstation in seconds for comparison. Digitization also allowed for business models that located credentialed radiologists in different countries and time zones to provide immediate interpretations of studies done anywhere in the world on a 24‐hour basis.

3. It allowed for new analyses of radiologic images.

This improved diagnostic accuracy and created new diagnostic applications. CT scans were the state‐of‐the art imaging technology pre-digital. But these required radiologists to hang multiple films in sequential order on large screens and slowly move through planes of the patient’s body to identify disease. Digital films could be viewed in “stack mode,” allowing the radiologist to page through a large set of images quite rapidly (critical as the number of images per study increased dramatically). A 3D representation could also be created from a sequence of images and rotated and viewed from different angles. Many other types of “digital‐assist” technologies have been developed and used to interpret both 2D imaging and a host of new 3-D acquisition technologies. None of this would have been possible in film‐based radiology.

4. It had a tangible impact on “quality.”

A little-remembered aspect of pre-digital radiology was the loss of large numbers of patients’ films forever – either misfiled in the library or checked out and never returned. Like in pathology, some radiologic findings can only be interpreted when a current study is compared with a prior one and a key change is spotted. It was not unusual, in busy medical centers, for 10 or 20 percent – or even more – of the films in a library to be lost, and the impacts on patient care are obvious.

What about WSI in the pathology setting?

Does it ultimately remove costs from the system?

Although it suffers from the same hardware, software and IT costs as digital radiology, WSI doesn’t eliminate any of the costs of slide‐based pathology, as going digital eliminated all the costs of film-based radiology, because WSI still requires the glass slide as a starting point. In fact, WSI increases some costs – for instance, WSI imagers’ focusing algorithms require a much better-quality glass slide than human pathologists, who can focus up and down or ignore preparation artifacts. The WSI imager’s rejection rate requires more recuts and sets the bar higher for slide quality. Use of whole-slide images also requires more processing power, storage capacity and bandwidth than digital radiology images – why? Because they’re in color, and because even “2D” images are actually 3D because of the focusing planes in a slice of tissue on a glass slide. These factors increase the cost hurdle for digital pathology. What about glass slide storage costs, often cited as a business justification for WSI? Unlike in radiology, the relative cost and space requirements of a decades‐long archive of glass pathology slides are actually quite manageable.

Will it improve the workflow or efficiency of a pathologist?

This is also a frequently‐heard business justification for investment in WSI. To even consider answering “yes” requires a “perfect world” of complete adoption – a leap from the current state in most clinical settings. What would that perfect world look like?

• All slides on all cases would be imaged.
• The imaging archive would be maintained on a pathology PACS system integrated into the anatomic pathology laboratory information system (AP‐LIS).
• Pathologists would need workstations with adequate processing power, high‐quality display capability and sophisticated human‐interface technology. The workstations must function like a microscope currently does – letting users rapidly view all parts of a slide at multiple magnifications and in multiple focusing planes.

But this perfect WSI system doesn’t exist, although it has long been the “Holy Grail” of otherwise very capable technology companies. Various halfway measures that have been advocated for WSI, like only imaging key slides (or only cancer cases, or only consultation cases, or only tumor board cases…), are ways to mitigate the cost – but because they still require glass slides, they’re inherently costly and inefficient, especially if the digital images aren’t completely integrated into the AP-LIS and require a separate program to view and manipulate.

Where can WSI improve pathology workflow and efficiency today? There are narrow applications where it’s already a reality – most notably, the remote interpretation of urgent cases or frozen sections that would otherwise require either the pathologist or the slides to travel. That’s especially useful for rural and specialty hospitals, or for cases that require subspecialty attention. Such productive uses of WSI improve the workflows of individual pathologists by sparing them long-distance or out-of-hours travel and ensuring that they can get (or give) answers in a timely manner. Although these applications of WSI contribute to the efficiency of the individual pathologist, because the proportion of such cases needing remote interpretation is small, the gains in efficiency aren’t comparable to those radiologists saw in their transition to digital.

Will WSI allow for new types of machine analysis of pathology slides?

In this area, WSI does indeed have promise. Various forms of image analysis are already available that take advantage of digitization. The most widely adopted example is the use of imagers to screen liquid‐based cervical cytology preps – truly a success story in the application of digital pathology imaging. Similar applications in histology might include identification of “rare events” in biopsies, such as cancer cells among benign cells, or an acid‐fast bacterium on a stained slide. Another real-world analytic application is in the scoring or quantification of characteristics on stained slides – for instance, the percentage and intensity of cells staining for ER/PR, or the proliferation index using Ki‐67-stained slides. Some researchers are even working on algorithms to allow software to learn from “viewing” digitally imaged pathology slides – with the goal that someday, computers might become artificial intelligence‐enabled diagnosticians capable of assisting or even replacing pathologists (“computer-aided diagnosis”).

Although image analysis is an exciting business justification for WSI, the range of applications for which it’s currently needed is small, and competing technologies like liquid biopsy, molecular techniques or in vivo microscopy are ultimately likely to advance faster, and be adopted more widely and cost‐effectively, than WSI.

Can WSI have a tangible impact on “quality” in pathology?

Here is an area where WSI has real‐world applications today, and one of the best business justifications for the technology. In areas like education, competency or quality assessment, and consultation on select slides or cases, WSI can be much more cost-effective than formats requiring transportation of a physical slide. WSI is already used to teach the younger, more “digital-friendly” generations in medical schools, and it’s expanding into residency training programs and continuing education for experienced pathologists. In competency assessment, too, digital images can be used to assess practitioner competency and quality in a standardized way – something Canadian pathologists have already pioneered on a provincial basis.

WSI can also be used for consultation on difficult cases. Many diagnostic concordance studies have shown that pathology is very much an interpretive and subjective specialty, one which is best practiced in a consultative environment – and WSI can facilitate rapid and simultaneous consultations from multiple colleagues. This application of WSI has not seen enough investment. There could be a few reasons – resistance from pathologists uncomfortable with viewing slides digitally; a lack of well-developed payment models; institutional IT limitations; or fear of legal consequences like malpractice, running afoul of patient privacy laws, jurisdictional licensing or other regulatory requirements. Despite its utility, the business case for this application has not been well-made. The costs and benefits have been poorly measured, defined or presented, and many times, the cost has been inflated by presenting unnecessarily expensive “all‐digital‐lite” plans.

Let’s take a broader view

While this comparative look at the adoption of digital radiology and pathology through the lens of the business case can be useful, there are a few key aspects not illustrated by the radiology–pathology construct.

The first is the broader definition of “digital pathology,” which includes modalities other than WSI. This is important because these technologies are much less expensive, more widely available, and often very much adequate to the task at hand. They include microscope-mounted digital cameras (or even smartphones!) whose images can be used for education, quality assurance, consultation, remote diagnosis or triage, or in communications between pathologists and clinicians to facilitate patient care. A “store and forward” model using a simple photomicrograph of a key slide can transform a time-consuming process (sequentially transporting a glass slide from one person to the next) into a time- and cost‐efficient consultation with multiple colleagues simultaneously that can be key to quickly and accurately diagnosing or triaging a tough case. Streaming video can also have real‐world uses – for instance, in the operating room, for frozen sections, or during regional tumor board meetings. In part due to the focus on WSI, the benefits of these lower-cost applications have not been fully realized, nor the workflows optimized. While WSI can also serve these needs, it’s much more expensive and the additional benefits may not be worthwhile.

A second aspect to consider is the possibility that fundamental changes in market structure, the legal and regulatory environment, the cost of WSI, or a breakthrough in technology may alter the cost-benefit calculation. What if a marked shortage of pathologists occurred in a region, or a large new market opened? What if regulatory or licensing barriers fell, so that foreign‐licensed pathologists or even non‐pathologists could read cases at significantly lower cost? What if the cost of WSI plummeted due to a technological advance – say, something that eliminated the need to produce a glass slide? Or there was a compelling breakthrough in artificial intelligence, machine learning or computer-aided diagnosis? Such events may seem unlikely, but could well change the calculus for the business case for a technology like WSI. The impact of a different environmental context on the case for WSI is illustrated by its accelerated adoption in some companies in limited research or in veterinary medicine settings (where the legal and regulatory barriers are lower, business uses differ and the pathologists may be fewer and farther between), as well as in unique clinical settings like rural and specialty hospitals (where external factors justify the necessary investment).

The reality check

Although WSI is now a “maturing” technology, there is not yet a compelling business case for wholesale deployment. However, WSI and other digital technologies are certainly of value, and I believe they will – and should – be adopted when the use case is appropriate. It is also prudent for pathologists to keep an eye out for potential “killer apps” or environmental changes that could upend traditional practice. But at the same time, it is important not to underestimate the efficiency of a pathologist looking at a slide on a microscope, or the cost and difficulty of radically changing that time-tested workflow using a single technology like WSI. Any such change requires a critical analysis of costs and benefits much like the one conducted in radiology – despite the not‐insignificant proportion of radiologists uninterested in “going digital,” the move made sense. That’s not currently true for WSI, something the market has validated by its slow uptake of clinical WSI over the past decade. It is important not to be overly wowed by a technology for its own sake. No matter how “cool” it is, one always must ask, “Is this a ‘solution in search of a problem,’ or a real game-changer?”

Luke Perkocha practices anatomic and clinical pathology, with a subspecialty in dermatopathology, at Kaiser-Permanente in Northern California. He has a degree in business administration and writes and speaks about business and technology/informatics topics in pathology and laboratory medicine.

Read more:

A Roadmap to the Future
By Alexi Baidoshvili

Part of a Larger Whole
By Marcial García Rojo

Techniques in Transition
Michael Schubert interviews David Snead

A Business Case for Common Sense
By Liron Pantanowitz