Digital pathology lags radiology in maturity, but offers potential payoff

Efforts to use of digitized images require more technical prowess, but could achieve a bigger payoff because of the importance of pathology in diagnosis and treatment, says Jeffrey Alan Golden, MD.

When it comes to digitizing images, pathology is lagging behind radiology—but with good reason, says Jeffrey Alan Golden, MD, chairman of the Department of Pathology at Brigham and Women’s Hospital and professor at Harvard Medical School.

Radiology has gotten a big head start on pathology, having converted to digital images more than 25 years ago, according to Golden. But he contends that pathology’s late adoption of digital imaging is the result of some practical and financial obstacles that put it at comparative disadvantage to its sister specialty.

“When radiology went to digitization, they saved money by eliminating the costs of film, developers, chemicals and storage space,” says Golden. “It also enhanced care because you were able to disseminate the images to where they were needed. Instead of that film being developed back in radiology and then having to be brought back to the intensive care unit, for example, as soon as it was imaged, it became digital and was available on multiple computer screens simultaneously.”

In an already financially and operationally stressed healthcare environment, he observes that digital radiology is the hands-down winner because digital pathology represents a huge added cost.

“Many, if not most, of the practical benefits realized by radiology would not be achieved with pathology digitization,” writes Golden in an editorial published earlier this month in the Journal of the American Medical Association. “An anatomic pathology workflow that includes digital pathology will not reduce or remove the need to produce and ultimately store glass slides of pathology specimens. Instead of any reductions, digital pathology will require additional workflows, personnel, equipment and, most importantly, storage of data.”

With regard to the latter, he contends that digital pathology images are at least 10 times larger files than radiology images—requiring substantial data management and storage capabilities. “Radiology is looking at resolutions in the maybe millimeter to sub-millimeter, but we look at resolutions that are down to microns,” Golden adds in describing the granularity of pathology images. “The density of data that you have to have on a single image to get that is extraordinary.”

At the same time, he boasts that about 70 percent of the data in electronic health records comes from pathology departments.

While digital pathology has made significant strides, it has not been broadly adopted, according to Golden. A major impediment to adoption, he says, has been the “lack of an FDA-approved methodology for being able to use digital pathology for primary diagnostics in a hospital.”

However, Golden says he is encouraged by the FDA’s approval in April of the Philips IntelliSite Pathology Solution, the first whole slide imaging (WSI) system that enables review and interpretation of digital surgical pathology slides prepared from biopsied tissue. The system enables pathologists to read tissue slides digitally to make diagnoses, rather than looking directly at a tissue sample mounted on a glass slide under a conventional light microscope.

Also See: New tech could change how pathologists view tissue samples

“There are things that are moving very quickly that will accelerate the adoption of digital pathology, because now it can be used—and we are using it,” adds Golden.

Nonetheless, he believes that radiology—as opposed to pathology—is “well positioned” to leverage artificial intelligence for diagnostics. “When you do this kind of artificial intelligence work, you need hundreds, thousands, tens of thousands of images to train the computer to be able to learn,” comments Golden.

Still, he sees tremendous progress in the area of artificial intelligence and pathology. His editorial in JAMA earlier this month, discusses a new study which found that an AI application could detect lymph node metastases in breast cancer patients as effectively as pathologists.

“While still requiring evaluation within a normal surgical pathology workflow, deep learning has the opportunity to assist pathologists by improving the efficiency of their work, standardizing quality and providing better prognostication,” writes Golden, who does not see pathologists being replaced by machines but as a complement. AI makes pathologists more efficient, he contends, and “potentially it improves the quality of what we do.”

According to Golden, another obstacle to pathology digitization is the lack of an agreed-upon standard for structuring the data. “That doesn’t quite exist yet—at least, not highly specified,” he says. “So, while radiology has a DICOM standard that allows for anyone to read images, that same standard doesn’t yet exist—or more precisely, it hasn’t been clearly defined. That’s going to be a major task going forward.”

Ultimately, Golden believes the challenge for digital pathology is to bring a true value proposition to the healthcare system before it is more widely adopted and viewed as a requirement for patient care.

“It’s a steeper slope for pathologists to climb than radiologists, but it’s one that we’ve started on and nobody is turning back,” concludes Golden. “It’s a very exciting area, and it’s one in which we still have a lot to figure out.”

More for you

Loading data for hdm_tax_topic #better-outcomes...