Most of the AI conversation in spine and pain practices centers on the back office: prior authorization, peer-to-peer scheduling and revenue cycle management. Nikhil Shetty, MD, wants to move it into the procedure room.
Dr. Shetty, MD, CEO of Midwest Interventional Spine Specialists, a single-specialty interventional pain practice and ASC based in Munster, Ind., is also the founder of Enhanced Fluoro, a company developing an AI device that interprets contrast patterns in real time during fluoroscopy-guided procedures.
“Clinical use for AI is somewhat underutilized,” Dr. Shetty told Becker’s.
The device is an edge unit that attaches to an existing fluoroscope, so practices don’t need to replace equipment to use it.
“There’s no added capital equipment costs,” Dr. Shetty said. “It’s an add-on device.”
As contrast spreads during a procedure, the unit reads the pattern and is designed to alert the physician to potentially dangerous spread, such as contrast entering a blood vessel or the intrathecal space. Dr. Shetty said it does so without disrupting the clinical workflow or adding time to the procedure.
Machine learning engineers spent the past 18 months building the algorithm, feeding it tens of thousands of fluoroscopy images. Many came from public sources; others were deidentified images from the practice. The goal was to teach it to tell a typical epidural contrast pattern from intravascular, intravenous and intrathecal spread.
The technology is still in early preclinical testing. Dr. Shetty’s center is running a 500-patient retrospective study, with IRB approval obtained in September, and is so far the only site involved. For now, the device runs in “shadow mode,” meaning physicians don’t see its reads during procedures. Researchers instead compare its interpretations with physician reads after the fact, checking whether the algorithm caught each hazardous pattern.
Dr. Shetty was candid about who stands to gain. Experienced pain physicians read contrast patterns through years of pattern recognition, he said, and for them the tool “really doesn’t do much.”
The value lies with fellows, newly practicing physicians and advanced practice providers. For them, the device acts as a safety backstop, flagging hazardous patterns early, before too much radiation or contrast is used and before a potentially disastrous outcome.
“Enhanced Fluoro is a safety tool,” he said. “It’s designed as a safety backstop for these types of procedures.”
That group may be doing more of this work in the years ahead. Dr. Shetty pointed to Northwest Indiana, where for at least the past decade myelograms have been performed by nurse practitioners or physician assistants, with a radiologist reading the images afterward. As interventional pain physicians shift toward more advanced minimally invasive spine procedures, he said, more injections could be referred to other physicians and advanced practice providers.
On the business side, the measurable return comes from less fluoroscopy time, which shortens OR time and reduces radiation and contrast exposure, Dr. Shetty said. The longer-term payoff is confidence: physicians who train with the device could carry it into practice as an assistant.
He expects the rollout to start with teaching universities and training programs, getting fellows familiar with contrast patterns, before broader adoption over the next two to three years.
The current version works only for procedures that use contrast, such as epidural injections. Radiofrequency ablation and spinal cord stimulation typically don’t use contrast.
Dr. Shetty stressed that the device is not meant to replace physician judgment or interfere with the patient-physician relationship. The physician still decides how to proceed, and the patient’s consent doesn’t change.
“If we are able to provide additional data points for clinical providers to make decisions in a more informed manner, then it’s done our job,” he said.
He sees Enhanced Fluoro as one of many clinical AI use cases set to expand over the next two to three years.
