From sophisticated surgical planning to staggeringly accurate navigation and medical imaging, artificial intelligence (AI) is shaping modern orthopedic technology in profound and complex ways. But while AI is undeniably paving the way for real advancements in orthopedics, obstacles prevent this technology from being more widely adopted. While the timeline for broader acceptance of AI in orthopedics is uncertain, it’s clear that the vast scope and power of this technology is already making an indelible impact.
How AI is Shaping Orthopedic Technology
Exponential growth in computational power is giving AI the ability to accomplish things in orthopedics that were recently thought impossible, said Peter Verrillo, Co-Founder and Chief Executive Officer of the surgery platform Enhatch. Before Verrillo launched his company, he was frustrated with the limitations of using CT data while engineering orthopedic implants.
“We needed more advanced computational tools,” Verrillo said, “so we pursued different technologies that would help us understand human anatomy.” Inspired by products like Apple’s iPhone, AI was the obvious answer for Verrillo because of its seemingly limitless potential. “Every five years, developers have about 100x more computational power available, so that opens up areas that we were never able to explore before.” Described by Verrillo as an “intelligent surgery platform,” Enhatch’s comprehensive, AI-driven system facilitates implant- and patient-specific instrument design and builds efficiencies into orthopedic surgeries.
AI’s ability to comb through vast assortments of data and develop insights based on what it observes is what makes it such a valuable asset in orthopedics. Complex and laborious tasks that used to take weeks are now being accomplished in seconds, Verrillo said.
Verrillo’s company is a prime example of the different ways that AI is being applied to orthopedic technology. The Enhatch platform streamlines tasks as diverse as surgical planning and navigation to implant and instrument design, manufacturing and logistics, intending to help orthopedic surgeons save time.
According to Verrillo, a sharp increase in orthopedic cases requiring preoperative planning is stretching surgeons thin. Enhatch leverages AI to help them get through the preoperative planning step faster. “If I have a proper preoperative plan, I can port that to any intraoperative solution,” Verrillo said. “Let’s take patient-specific instruments, for example. AI constructs the surgery plan and generates patient-specific instruments to perform the surgery. Doing this alone can take five to ten minutes off of the procedure.” Verrillo claims that Enhatch saves orthopedic surgeons up to 40 minutes of preoperative planning.
Using AI, Israeli startup Zebra Medical Vision addresses the insatiable demand for healthcare imaging. According to Clinical Director Orit Wimpfheimer, M.D., a swelling international middle class combined with aging baby boomers in need of medical care is driving sharp increases for radiological expertise and outpacing supply. “Medical imaging technology has advanced, enabling more detailed and complex imaging, which significantly improves patient care, but at the same time creates an enormous workload for the radiologist,” Dr. Wimpfheimer said.
Zebra Medical’s AI-driven technology scans a vast library of over 30 billion imaging studies. Then, the company’s data scientists and engineers comb through the findings to construct algorithms for various medical imaging modalities that are “fine-tuned to the individual needs in each of those domains,” Dr. Wimpfheimer said. “Medical imaging AI has two key roles. The first is to assist the radiologist to be more efficient, effective and accurate. The second is to harness the vast amount of data created by medical imaging to provide new and improved tools for clinicians to treat their patients.”
When applied to orthopedics, Zebra Medical’s platform aims to provide an improved method for converting 2D x-rays into 3D images. “Currently, this can only be done using a CT scanner, which is non-portable, expensive and time-consuming. AI-derived 3D images, accurate measurements and bony landmarks would be immensely beneficial to the orthopedic surgeon,” Dr. Wimpfheimer said.
Zebra Medical’s orthopedic offering is currently in development, with the first stage pending with FDA. In late 2019, the company entered a global co-development and commercialization agreement with DePuy Synthes, which plans to incorporate Zebra Medical’s technology into its VELYS Digital Surgery solutions offering for pre-op, operative and post-op patient care.
“Surgery is tremendously complex,” said Gabriel Jones, Co-Founder and Chief Executive Officer of the surgical navigation company Proprio. The company’s platform is currently conducting pilot orthopedic surgeries for testing purposes. When asked how their AI and Machine Learning technology enhances surgical navigation, Jones explained how Proprio’s tech captures wide swaths of valuable data during surgeries and processes them in ways that guide surgeons and build efficiencies in real time.
“AI and Machine Learning are powerful tools we leverage to parse and process many data streams so surgeons can make vastly more informed decisions,” he said. Jones cited a frustration of surgeons of having to look for crucial information from multiple sources during procedures. “Our platform will harmonize that information and allow the surgeon to filter it so they get only what they want when they want it.
“The best technologies are not ‘in your face.’ They melt into the background and deliver an experience that is both subtle and transformative,” Jones continued. “A tremendous amount of costly equipment is needed today to perform orthopedic surgeries. Setting it up, often numerous times for one case, is costly, time-consuming and frustrating. Our platform will reduce, or replace entirely, this equipment and its burden on efficiency.”
The Salt Lake City-based orthopedic technology company OrthoGrid has pioneered a self-described “AI-driven ecosystem” that detects and tracks specific anatomical landmarks through a sophisticated grid network. “The capability facilitates consistent, accurate grid placements throughout the procedure to help surgeons achieve the desired anatomic and implant alignment,” said Joshua Cates, Ph.D., Vice President of Research & Development at OrthoGrid. “This is enabled by AI that is designed to minimize direct surgeon and/or operating room staff contact or interaction with our software to enhance surgeons’ workflows and helps surgeons achieve reproducible results.” Citing OrthoGrid’s level pelvis tool as an example, Dr. Cates claims that the technology behind the PhantomMSK Hip product gives surgeons the ability to maintain proper views of the pelvis throughout direct anterior total hip replacement procedures in a hands-free fashion.
In addition to mapping anatomical landmarks, PhantomMSK Hip provides a distortion correction tool engineered with “a smart algorithm enabled by detection of dots on a calibration array in the fluoroscopic field and distortion-field computation,” Dr. Cates said. “We are currently developing a next-generation distortion-correction algorithm that uses a deep learning model for calibration array dot detection, which is made possible by analysis and computer learning from thousands of intraoperative and simulated fluoroscopy images.”
To illustrate the benefits of OrthoGrid’s technology in surgical contexts, Dr. Cates referred to a study of patients conducted between July 2018 and January 2020. Each underwent unilateral total hip replacement via the direct anterior approach that was performed by a fellowship-trained orthopedic surgeon. According to Dr. Cates, “The surgeon, who used PhantomMSK Hip (Phantom) to assist in the positioning of total hip arthroplasty components, found that consistent accuracy with the Phantom was achieved while maintaining a low overall surgical time and fluoroscopic exposure. This study measured leg length discrepancy, global hip offset and acetabular cup abduction angle, along with surgical time and fluoroscopic time.” The study’s findings were published online earlier this year by the Journal of Arthroplasty.
While some AI technologies help orthopedic surgeons build efficiencies during procedures, others give clinicians crucial and unprecedented insights into how patients are healing from traumatic injuries. Stephen F. Badylak, DVM, Ph.D., M.D., Deputy Director and Professor of Surgery at the McGowan Institute for Regenerative Medicine at University of Pittsburgh, is heading a team that’s developing AI-driven technologies that monitor patient healing. “In the current work we are conducting, bio-electrode technology capable of measuring almost any biomarker (such as a proinflammatory cytokine, or an anti-inflammatory cytokine) in realtime delivers this information to an AI device that can utilize existing or newly developed databases that inform the clinician of the state of healing,” Dr. Badylak said.
Through smart bandages that monitor the wound site over time, clinicians can determine whether it’s healing correctly. “One can also treat the wound site with selected signaling molecules that alter the course of healing such that a regenerative response is induced rather than a proinflammatory scar tissue response,” said Dr. Badylak. He noted the platform he’s developing can be used to avoid adverse outcomes after injuries and surgeries by giving physicians the power to intervene early to promote healing. This technology is in the R&D phase but on the “fast-track to clinical translation,” he said.
Liane Teplitsky, Vice President and General Manager, Worldwide Robotics at Zimmer Biomet, cited AI as a crucial element of the company’s current and future operations. “AI plays a critical role today, and most importantly in the future, of a product we launched in the spring of this year called OrthoIntel.” Described as an “orthopedic intelligence platform,” OrthoIntel gathers data and clinical insights from patients pre- and post-operatively in tandem with intraoperative information to create reports with metrics designed to give knee and hip replacement patients better recovery outcomes. The data is collected via Zimmer Biomet technologies like mymobility and ROSA Knee. “The goal is to help surgeons improve outcomes and patient satisfaction.”
What’s Holding AI Back in Orthopedics?
Until recently, an outdated and ill-fitting regulatory framework made it difficult or even impossible for some orthopedic technologies to reach premarket clearances, De Novo classifications and premarket approval statuses. FDA guidelines favor so-called “locked” AI technologies that require humans to update the technology, over tools that adapt by learning and changing over time.
FDA has made progress in recent years to address its antiquated AI-related regulations and is currently considering a total product lifecycle-based regulatory framework that would support the ever-evolving nature of modern AI technologies in orthopedics and throughout healthcare.
A recent announcement from FDA explained, “The FDA is considering a total product lifecycle-based regulatory framework for these technologies that would allow for modifications to be made from real-world learning and adaptation, while still ensuring that the safety and effectiveness of the software as a medical device is maintained.”
While FDA continues to decide how to modernize its approach to AI-integrated healthcare technologies, orthopedic companies cite other significant challenges. OrthoGrid’s Dr. Cates asked, “How can a surgeon trust an algorithm result if they don’t understand how the system derived the output or solution?” While Dr. Cates believes OrthoGrid’s platform is careful to let surgeons know they’re always in control, he acknowledged that the so-called “black box” issue is a significant challenge for AI in general. “Our software does not attempt to make decisions for the surgeon.”
Jones of Proprio believes a major challenge for AI in orthopedics are companies that build equipment and devices that aren’t designed to keep up with the technology. “Throwing AI at an existing database of healthcare data can only succeed to a very limited extent,” he said. “To build a truly AI-driven company in medicine, you must build the company as a hybrid of the very best technical and medical device talent. We think that is a core difference of what we are doing at Proprio compared to the rest of the industry, which still takes a very mechanical and electrical engineering approach to development.”
What the Future Holds for AI in Orthopedics
“The potential of AI in orthopedics is virtually unlimited and will ultimately lead to a better surgical experience for the patient,” said Zimmer Biomet’s Teplitsky. “It will allow them to participate in their own care in new and exciting ways and will offer choices that were previously unavailable to them. For surgeons, AI will uncover new insights not previously seen and allow them to leverage data in order to make informed decisions and optimize care.”
Zebra Medical’s Dr. Wimpfheimer shares Teplitsky’s optimism regarding the prospect of AI being used to profoundly benefit orthopedic patients. She foresees a vision for the future for orthopedics and other healthcare fields in which AI transforms immense sets of data into powerful workflow tools for medical experts.
“There are still a lot of people who feel like [AI is] always going to stay the same,” said Enhatch’s Verrillo. “But when we look at AI, 3D printing and the miniaturization of sensors, it all adds up to a more data-driven world. Many of the folks running orthopedic organizations today are just not used to that. They’ve got metal, plastic and sales reps. If you’re living in that world, five years from now you may not be in business. The companies that take advantage of what’s possible now will be leading the pack later.” In 2019, Verrillo wrote an article envisioning a future where AI, Machine Learning and 3D printing technologies produce bones and other body parts to replace the ones that fail in patients. “Enhatch is ushering in an era where death is optional,” reads a line from the article.
Dr. Cates has a markedly more modest take on the future of AI’s relationship with orthopedic technology. “I don’t believe we are at the point of self-driving cars in most areas of healthcare,” he said. “For example, I don’t believe an autonomous robot is going to be realistic or desirable in orthopedic surgery any time soon.”
Dr. Cates noted that AI systems that give assistance and guidance to orthopedic surgeons in real-time is a more realistic vision of the future than autonomous AI-driven robots that perform surgeries. “To extend the analogy of AI in automobiles, what is going to have an immediate impact in orthopedic surgery are the guidance and warning systems––the sensor that detects a car in your blind spot before you change lanes, the automatic braking assist when you’re too close to the car in front of you, the tire pressure warning system, etc. What can be deployed today for orthopedic surgery are the computer vision sensors and smart tools that bring additional information and insight to the surgeon in real time and provide guard rails for a procedure to help the surgeon prevent mistakes or avoid suboptimal surgical outcomes.”
(Image courtesy of Proprio.)
Patrick McGuire is an ORTHOWORLD Contributor.