A Path to Follow?
By Kathy Hardy
Vol. 18 No. 12 P. 20
Radiology and pathology are complementary disciplines. Technology may (or may not) bring them closer together.
Radiology and pathology reside at the center of disease diagnosis. However, the workflows of both specialties occur in separate silos with no connection between the departments or their reporting systems. Data gathered by each specialty are essential to making proper diagnoses and determining patient management and treatment options. While developments in digital pathology imaging could bring these parallel worlds together, there is still significant distance to travel.
Unlike radiology, where technology advances in digital imaging have proliferated, digital pathology has not progressed at the same pace. Just this year, the FDA granted permission to Philips to market its IntelliSite Pathology Solution for primary diagnosis, the first whole slide imaging system that allows for review and interpretation of digital surgical pathology slides prepared from biopsied tissue. This is the first time the FDA has permitted the marketing of a whole slide imaging system for these purposes. The system scans and digitizes surgical pathology glass slides prepared from biopsied tissue.
"Just as radiology became digital some 20 years ago, histopathology is poised to do similarly," says Homer Pien, Philips' chief technology officer for imaging. "When radiology became digital, remote consultations and image sharing became possible, teleradiology increased the availability of rapid radiologic diagnosis, and the development of image processing and visualization techniques dramatically improved the efficacy of radiology. Digital pathology will surely go the same route."
Pien notes that as artificial intelligence becomes more widely adopted in health care, the roles of radiologists and pathologists could be redefined. These positions may ultimately merge to create "diagnosticians" with deep knowledge of disease processes who may have greater insights and expertise in both diagnoses and treatment pathways. Radiologists and pathologists have varying views, however, on the value of digitized pathology images and the role this technology could play in a potential convergence of these medical specialties.
Digitization and Convergence
Christopher Garcia, MD, director of pathology informatics at the Medical College of Wisconsin, focuses his research on the digital transformation of medical diagnosis and the emergence and intersection of digital imaging, machine learning, decision support, and health care delivery. He recognizes certain potential benefits of a convergence of radiology and pathology, including standardized reporting, the efficiency of accessing reports via a single portal such as PACS, the development of a registry of diagnostic imaging and pathology data into one resource, tumor board presentation, and research and innovation. All that said, just one factor could be the ultimate catalyst of pathology and radiology convergence.
"When we find the benefit for improved patient care, that's the factor that will drive the possible convergence," Garcia says. "For now, we just need to get better at working with each other, with the goal of showing improvements in quality. This is especially important with value-based care."
The path to convergence could be a matter of accepting a paradigm shift in the way doctors see the two specialties. As University of Pittsburgh Medical Center (UPMC) Director of Radiology Informatics Christopher Deible, MD, explains, radiology's role in patient diagnosis is often as a guide along the course of the overall treatment process. Pathology usually involves a more targeted singular event for a disease process.
"When considering a patient's treatment, a doctor can look at radiology images and interpret the images in the context of a patient's illness," Deible says. "Radiology could show that a biopsy is necessary, which is where pathology comes in. However, the pathology report will result in determining the exact course of treatment. If the two specialties could reach a point, through technological advances, where they provide similar results, that could be a driver to a convergence."
For now, Deible says the use of digitized pathology images should be considered on a case-by-case basis. For example, digital images of pathology slides can be shared via mobile devices for use in remote locations or to dovetail with teleradiology efforts. However, he sees this occurring in more high-specialty medical situations rather than for widespread adoption.
"Pathologists will look at radiology images, and radiologists have a need to work with pathologists and review pathology results for patient follow-up," Deible says. "That process speaks to patient quality of care but isn't necessarily a driver for the digitization of pathology images."
According to Pien, the challenges for adopting digital pathology are analogous to the adoption of digital radiology. The use of digital imaging will require labs to make time and technology investments and pathologists to get comfortable with workflow changes. Due to the size of pathology images, hospitals will need to make sure the IT infrastructure has sufficient bandwidth to handle the data transfer. Image processing techniques will provide enhancements to digitized data but will require time to adapt. Many in the field contend that the biggest challenge in digitizing pathology is the time involved in developing an efficient technological process for converting samples on slides to digital images.
"This is more about the technology and the challenge of converting to digital pathology," says Brian J. Kolowitz, DSC, MSC, MBA, director and principal architect of advanced medical imaging for UPMC Enterprises. "Pathology departments can't plan workflow transitions to digital while waiting for technology. Everyone generally sees value in digitized pathology, but it's hard to sell the organization on a three-year plan to go digital while you're waiting to see what technology is approved."
Another challenge, according to Kolowitz, is the cost of storing all of the prepared glass slides. "When radiology transitioned to digital, they phased out film storage," he says. "Even though digitized pathology enables us to share images, we would still want to save the tissue samples that are on the slides."
File size is also a factor. There's a large amount of data required when storing slides digitally. According to Garcia, pathology images can be 0.7 GB to 7 GB per slide, and one case can include 50 slides.
"It can be cost prohibitive," Kolowitz says. "In pathology, you don't have the big value proposition on the back end as you did with radiology."
With radiology and pathology working in different silos, there's a communication gap in the flow of information, Kolowitz says. Feedback isn't optimized or as efficient as it could be, in regard to discussing patient diagnoses. Once pathology images are digitized, radiologists and pathologists gain value from being able to view images. This capability positively impacts diagnosis and patient satisfaction.
On the physician side, Liron Pantanowitz, MD, UPMC's director of pathology informatics, says that there is a greater volume of requests for radiology images than for radiologists' reports about the images. Not so in pathology, where, he says, reports carry more weight than images.
"Doctors can look at radiology images and see anatomical structures like bones and tissue," he says. "With pathology, reports detailing what's found on the slides carry more weight with doctors than actual images of samples on the slides."
Radiology has the DICOM standard for handling, storing, printing, and transmitting information. There is no such protocol for digitized pathology images, and Kolowitz says pathologists might want to keep it that way.
"The concern with applying standards like DICOM to digitized pathology images is that this is an emerging and evolving area," he says. "While standardization would help to consolidate things like file formatting, it could impede the flow of information."
Kolowitz adds that standardization could help with interoperability within the specialty but may compromise other factors. "Data analysis can be compromised if the standards prohibit our ability to query and leverage the data," he says.
"Standardization could also impact how you search data. For instance, there's no good way to search many PACS for patient outcomes. I think it's just too soon to bank entirely on a particular standard."
As Pantanowitz explains, the digitizing systems in pathology are not integrated as they are in radiology. In pathology, each step is proprietary. In addition, digital pathology systems don't incorporate the same functionality that radiology does in a PACS.
"For example, radiologists can report through PACS," Pantanowitz says. "In pathology, you can't do much reporting. Everything in the digital pathology process is standalone and doesn't incorporate into an ecosystem like PACS."
In addition, there is still cultural resistance to changing pathology workflow to allow for the digitization of slides. Radiology's transition to digital images started at the beginning of the process, with the digital capturing of images.
"In pathology, you're creating an additional step," Pantanowitz says. "You go from patient to slide to digitization to the pathologist. That additional step can be a disruption to workflow." Another workflow variation between the specialties is that pathology is a batch-related process, while in radiology, digital imaging involves one continuous workflow.
Learn From the Past
Transitioning from slides to digital images isn't as simple as following in radiology's footsteps, says Paul Chang, MD, FSIIM, vice chair of radiology informatics and former medical director of pathology informatics and enterprise imaging at the University of Chicago. As he explains, "You can't just dust off a radiology PACS and make a pathology PACS." However, he does see opportunity for pathology to learn from radiology's mistakes. Unfortunately, he thinks this is where pathology has missed the mark.
"Pathology could avoid the errors of radiology in making this transition possible," he says. "But they're not doing that. They are not learning from our mistakes."
Instead of focusing solely on the end product, Chang suggests that pathology apply informatics to workflow and interoperability as well. "They're not seeing all the benefits," Chang says. "It's not just about putting samples into a digital format. They should be using IT to embrace an optimized workflow. That's where radiologists started seeing improvements."
Procedural changes such as IT barcoding of samples could help, Chang adds, and this would occur before sample digitization. He also suggests a change to continuous inline processing of samples.
"Tracking inline workflow rather than batching would result in demonstrative value," he says. "However, adoption of this type of process would require a cultural change. Radiologists didn't see value in digitizing images until they moved beyond just digitizing images and focused on the workflow."
Comparing the adoption of digital pathology with that of digital imaging in radiology, Bruce Friedman, MD, emeritus professor of informatics in the department of pathology at the University of Michigan Medical School, says that radiology made quick business of deploying digital technology, with major benefits precipitating the change. Notably, the new imaging modalities in radiology were digital from birth, and, in many ways, were superior to older imaging technologies. In addition, the new modalities resulted in high profit margins, making them good business decisions.
Friedman believes that pathology could benefit from making images portable, making them available for automated image analysis and also for physicians seeking remote consultations.
These two factors could result in reduced costs and increased quality for the interpretation of pathology images, Friedman says.
Friedman also believes that adopting a radiology perspective regarding integrated diagnosis would help enhance pathology's adoption of digital images and, with that, a potential move to convergence with radiology.
"Radiologists understand the benefits of total automation," he says.
As technology makes its way from the imaging center to the lab, radiologists and pathologists could be sharing more than test results as they work toward convergent patient care.
— Kathy Hardy is a freelance writer based in Phoenixville, Pennsylvania. She is a frequent contributor to Radiology Today.