By Matt Skoufalos
Ever since its development was funded by the popularity of The Beatles (or the sustained research contributions of the British government, take your pick) medical imaging technologies have touched innumerable lives and become indispensable to the practice of health care. As imaging and the access to it continues to be the window into which medicine views not only its patients but, in a way, its entire practice, the iterations and influences that will shape the next wave of its development are not long in coming.
Neurosurgeon Neal Kassell believes focused ultrasound is one such technology on the threshold of a major leap forward, having expanded the number of mechanisms whereby it can affect tissue from three to 18 in the past decade. Kassell, founder of the Focused Ultrasound Foundation, calls it “a platform technology” with the potential to treat more than 70 known conditions, including brain tumors, Parkinson’s and Alzheimer’s diseases, epilepsy, and depression.
“This is a terrific example of the biological effects of physical energy, and it has the potential to be as revolutionary to therapy as MR scanning was,” he said.
It works by focusing multiple beams of energy through an acoustic lens, targeting a point deep in the body “with extreme precision and accuracy,” while sparing the adjacent, normal tissue, Kassell said. Focused ultrasound can destroy tissue, disrupt cell membranes, or even deliver high concentrations of pharmaceuticals with greater accuracy than intravenous or oral mechanisms. It can even reduce dramatically the systemic complications of other forms of radiotherapy, thereby enhancing the body’s immune response and augmenting the effectiveness of immuno-oncology drugs, Kassell said.
“People have become very excited about the fact that this is a totally noninvasive, therapeutic technology,” Kassell said. “It fulfills the criteria for the Holy Grail of modern therapeutic technology in that it improves outcomes and simultaneously decreases costs, if the potential is fully realized. It could be used to treat millions worldwide.”
Since the advent of focused ultrasound, more than 30 manufacturers have entered the marketplace – up from five as recently as 10 years ago – and 175 global research sites are pursuing various technical, preclinical, and clinical trials. As the cost of the technology continues to decrease, the industry is reaching “a critical mass of evidence” that found its tipping point in tumor research, and could expand to as many as 10,000 sites worldwide, Kassell said.
“In fields like biomedical research and medical device development, progress occurs exponentially, and we’re right at the inflection point of the curve,” Kassell said. “It’s still called medicine’s best-kept secret.”
Another of the leading trends in medical imaging relates to the outgrowth of artificial intelligence (AI) and machine learning, said Tomer Levy, General Manager, Workflow and Infrastructure at McKesson of Richmond, British Columbia, Canada. Although he regards AI as “still an open question,” the answers to which have yet to emerge, Levy said much of the hype around the technology is centered on realizing its full automation. To Levy, the Watson cognitive system from IBM demonstrates how the company has set itself apart in the marketplace, not only through its pioneering approach to AI development, but also in its formation of partnerships with vendors positioned to design a variety of applications for its technology. Even subsequent to its acquisition of Merge Healthcare, IBM has made its source code open and available for any vendor to incorporate into its own solutions. Levy said that strategy reflects the company’s maintenance of a commitment to advancing the entire industry.
“If Watson were only going through Merge, that limits the adoption of the technology,” he said, “but by opening it to other vendors, it can reach more [of an audience].”
A similar approach is being pushed by Zebra Medical Vision of Shefayim, Israel, a machine-learning, imaging analytics vendor that in November 2016 launched Profound, a web-based service that allows patients to upload their imaging studies for analysis by its algorithms. So far, the service is operating in Asia, Latin America, Europe, and the Pacific Rim, and the company says Profound can analyze cases of osteoporosis, compression fractures, fatty liver, coronary calcium, emphysema, and aortic aneurysms. But the service is a patient-facing companion to consultations that can help cut back on the wait times and possibly reduce anxiety among those awaiting a professional diagnosis.
“It’s not fully automatic and it will not replace a second opinion, but they have a few algorithms, and they will be able to run those algorithms on your image,” Levy said. “It goes along with the trend of consumerism and allowing patients to have access to their data.”
Levy also foresees greater attention to enterprise-class imaging needs, from data storage, retrieval, and access, to cybersecurity and workflow. Data systems vendors are pressed to design their products for interoperability as well as for information portability and patient privacy, and how capably they are able to do so will have a dramatic influence on the market, he said.
“Enterprise imaging is taking different shapes and forms; different providers use different parts of enterprise imaging to solve different problems,” Levy said. “As those diverse business strands are further integrated, we’re dependent on shifting from one side to another. Vendors need to align, but even if some vendors make a commitment to adopt the standards, we see the actual adoption in the market going low and slow.”
Enterprise PACS solutions offer a one-system-meets-all-needs approach, but a single database doesn’t offer enough flexibility in a mergers- and acquisitions-focused climate. When new facilities come online, a monolithic solution can slow full integration. On the other side of the fence, “a completely deconstructed PACS,” offers best-of-breed performance but raises the challenge of creating system-wide consistency and economies of scale when dealing with multiple vendors,” Levy said.
“I think now the industry is changing; looking for something in between,” he said. “Providers do prefer to have a single imaging [vendor] partner, but interoperability is a strategy or a goal. They may consolidate around a VMA or a single analytics solution but allow flexibility to sub-business units.”
The market for workflow solutions among enterprise imaging customers is another area of business Levy expects will evolve as OEMs work to balance interoperability and best-of-breed considerations. Centralized image storage and access demands mean that much of the focus on volume issues relies on efficiency.
“A lot of potential for quality of care is leading from workflow,” he said. “How do you decide who gets to read which study, and how do you prioritize these studies on the physicians’ workspace? If I make sure that the images are being read by the most qualified physicians in my entire enterprise, I can combine qualities and efficiencies that I couldn’t before.”
Levy foresees the imaging industry “trending very strongly towards two giant islands of information,” as niche providers turn to electronic medical records (EMR) management to resolve their data access and integration issues “for anything that is not imaging.” These “mega-VNAs” (vendor-neutral archives), which store imaging information enterprise-wide, are turning data itself into a platform, he said. When interoperability opens the door to innovation, a variety of companies will add analytical value to the handling and management of that data, and machine learning will both accelerate and complicate the outgrowth of this process.
“Interoperability is key, and a challenge to innovation if it’s not adopted quickly,” Levy said.
For Pablo R. Ros, professor and chairman of the radiology department at Case Western Reserve University and the University Hospitals of Cleveland Medical Center, workflow is certainly a linchpin consideration in the future of medical imaging. As the industry turns its eye toward managing, interpreting, and integrating the various streams of information comprising imaging data into better patient outcomes, the role of the clinician becomes one of absorbing and processing the depth of information available to him or her. The emergent field of radiomics, which maps imaging findings to laboratory results and genetic profiles, will further support the advancement of this integrated diagnosis, Ros said.
“One of the potentials for the future is to have a radiologist or a diagnostician that will orchestrate all of these three areas, and basically say, ‘Give me the patient and I will give you perhaps the most tailored possible therapy,’” he said.
Ros doesn’t think the same appetite for advancement applies as much to the market for scanning devices themselves; in a world where “nobody really has high-end money anymore,” he said the industry “is beginning to produce very solid workhorses” for the American market, just as it has done in those of emerging countries. Where manufacturers will make up the difference is in continuing to hybridize devices to facilitate high-end, multimodal applications, and adding complementary imaging techniques to other areas of health care, like the operating room.
“We still produce the Ferraris, but we produce more reliable, good, solid cars that don’t need much repairs,” he said.
Equipment spending isn’t only down, but radiologists are being pushed to do more with less, and for less money. When the onus on health care professionals is more on “first-time right” diagnoses, manufacturers are responding with technologies that better facilitate decision support and overall efficiency of process, said Rob Cascella, CEO of Diagnosis and Treatment Business at Philips.
“It puts a lot of pressure on equipment to communicate,” Cascella said. “The modality will really be taken for granted at some point. It will be the applications on the modality, the advanced acquisition, and the streamlined workflow, that will have to exist.”
Cascella agrees with Ros that iterative advances in high-end equipment won’t be as attractive to purchasers as systems that support post-image-acquisition services and which create a value chain that make the experience “more predictable” to patients, especially in hub-and-spoke health networks.
“I think the whole out-of-hospital market is thriving for things like that,” Cascella said. “They want a gantry that works every day that is reliable and easy to use. Where the investment comes into play is going to be what you use with this daisy-chain of gantries to help your team of radiologists become much more efficient and more clinically appropriate.”
That doesn’t mean that Philips isn’t working on advancing the technological sophistication of its gantries. Cascella said the company is placing a focus on “creating a gantry that changes the level of activity,” whether in terms of better dose management, more efficient operation, or better disease quantification – basically expanding its functionality to meet the demands of the procedures it performs while improving workflow at the same time it improves detection and disease quantification capabilities.
“A significant amount of R&D is going to how can we make equipment operate more efficiently,” Cascella said. “There’s [also] this whole notion of broader applications that are very disease-specific; a tremendous amount in that arena to make procedures easier to do with greater standardization and accuracy.”
At the intersection of these specialties, Cascella agrees with Levy that the incorporation of AI within the interpretation and interrogation of images will aid the decision-support focus, using deep machine learning to improve disease recognition algorithms, and “turn[ing] data into wisdom.” Integrating computer-aided diagnostics with pathology data, imaging data, and genomic informatics allows providers to offer more precise personalized medicine; the radiomics of which Ros spoke.
“We’ve developed a library of genomic algorithms, and there’s a bunch of informatics on the operating pathway side,” Cascella said, “while the other side of the shop is working on all the things to help the clinician get better at what they’re doing. This is so evolutionary that we’re right at the beginnings of this process.”
The ultimate terminus of the conversation ends with the next fundamental question, which Cascella said is eventually posed by every radiologist with whom he interacts: “When is the machine going to replace me?” He says that if medical devices can be leveraged to “wring out the efficiencies of a radiology department,” then the role of a clinician changes, but it isn’t eliminated altogether. The more interesting question that follows is, “When does the machine start to practice medicine?”
“No matter what the machine does, the radiologist should still provide the reason in terms of the interpretation,” Cascella said. “I think we have a long way to go to have the machine standardize what the radiologist is doing.”