Korea’s hospitals are rethinking how much contrast they really need.

As AI-powered image reconstruction begins to trim dosing requirements, suppliers such as GE HealthCare, Philips, and Bayer are recalibrating their bets on a market once defined by chemistry and now increasingly driven by algorithms.

The conversation has shifted since the European Society of Cardiology’s 2024 update, which elevated coronary CT angiography to a first-line test for suspected heart disease. The change expanded use of X-ray contrast media even as other global bodies tightened warnings for patients with kidney impairment or concern about gadolinium retention in MRI.

“The problem isn’t the agents, it’s how we use them,” said Park Hyo-jung, professor of radiology at Asan Medical Center, in an interview with Korea Biomedical Review. “Protocols need to adapt to patient risk rather than simply cutting dose for everyone.”

Park said newer guidelines increasingly emphasize substitution over premedication for patients who had prior allergic reactions. “Switching agents is the most evidence-based way to prevent recurrence, but it raises a practical question: switch to what?” she said. “We’re now studying the chemical side chains of iodinated agents to learn which alternatives are safest. If we can prove that, the impact would be huge.”

At Asan, she added, radiologists routinely consult allergy specialists for patients who experienced severe reactions. Skin tests help identify a safe substitute, and those results are stored in electronic records to guide future scans. “It’s a small system change that directly affects patient safety,” she said.

Manufacturers stake out their ground

GE HealthCare expects its iso-osmolar Visipaque (iodixanol), the only contrast approved by the FDA for coronary CT angiography, to gain new traction as Korea’s hospitals align with international standards. Omnipaque (iohexol), its lower-osmolar mainstay, is used in more than 200 domestic sites, according to the company. A recent meta-analysis found its ingredient iohexol associated with fewer acute hypersensitivity reactions than rival formulations, the company added. 

In MRI, GE points to Clariscan (gadoterate meglumine), a macrocyclic ionic agent designed for long-term stability. The company reports no confirmed cases of nephrogenic systemic fibrosis since its launch, citing two decades of post-market surveillance.

Bayer has centered its argument on safety data. Its iodinated agent Ultravist (iopromide) was used in the Lancet-published AMACING trial, which found that moderate-risk kidney patients did not benefit from prophylactic hydration before CT scans. The finding reshaped European kidney-risk thresholds in 2018 and reduced the number of patients classified as high risk, the company said.

“The agent’s low viscosity and osmolarity made that possible,” Bayer said in a note to Korea Biomedical Review, calling the trial a benchmark for safety-driven design.

Bayer’s Korean portfolio spans Ultravist, Gadovist (gadobutrol), and Primovist (disodium gadoxetate), plus the Centargo injector that automates saline priming and barcode documentation. The company says Centargo can shorten setup time by up to 60 percent while tracing each injection for audit compliance.

AI reframes the dose economy

For GE and Philips, the next advantage lies in software rather than chemistry. GE’s AIR Recon DL and Precision DL systems use deep learning algorithms to rebuild medical images from raw scanner data, reducing the grainy “noise” that can blur fine detail.

The technology, now built into the company’s newest CT and MRI scanners, lets hospitals either shorten scan times by half or use smaller doses of contrast dye while keeping image quality intact. “The goal isn’t to eliminate contrast,” the company said. “It’s to make every milliliter count.”

Philips Korea has taken a similar approach. Its SmartSpeed MRI and Precise Suite CT software use artificial intelligence to clean and sharpen images after acquisition, allowing radiologists to see vessels and organs clearly even when 30 to 40 percent less contrast agent is used. “We’re not replacing agents,” a company official said. “We’re extending their diagnostic life.”

Choi Joon-il, policy research director of the Korean Society of Radiology and a professor of radiology at Seoul St. Mary’s Hospital, said the technology shift is already changing how hospitals plan their purchases. “We are being asked to do more with less: less contrast, less radiation, less error,” he said in an interview. “AI can offset the loss of image sharpness when you cut dose, but it also raises questions about cost and liability if the results aren’t diagnostic.”

He noted that CT contrast use remains dominated by six non-ionic agents that are chemically similar and largely interchangeable. “Clinically, differences are minimal,” he said. “The real competition is in formulation tweaks and service. Domestic firms promote lower-concentration products and aggressive research support, while multinationals emphasize data and reliability.”

Hospitals weigh safety against economics

Both Park and Choi agreed that Korea’s dense imaging volume gives it an advantage in research but magnifies cost pressure. “Reducing contrast is good for safety, but smaller hospitals that can’t afford AI upgrades may fall behind,” Professor Park said. “It’s creating a technology gap inside the same healthcare system.”

Choi added that per-patient doses may decline as AI matures, but total usage will likely rise with the number of scans. “The market won’t shrink,” he said. “We are simply shifting from chemistry to computation.”

Not every company is following the AI wave. In September, Bayer confirmed it will postpone the rollout of its Calantic AI imaging platform in Korea, citing “market insights, technical considerations, and customer feedback.” The company said it will redirect those resources toward initiatives with clearer clinical impact while continuing digital integration across its injector systems.

The Ministry of Food and Drug Safety has cleared few new contrast agents in recent years, leaving hospitals reliant on multinational suppliers. Domestic producers focus on generics such as iohexol and iopamidol, competing through concentration options and GMP certifications rather than novel chemistry.

Park believes the next breakthroughs will come from organ-specific imaging. “Primovist was revolutionary because liver cells actively take it up,” she said. “Another agent like that would change practice again, especially in regions with high liver-disease burden.”

Both experts agree that the era of big chemical innovation in CT contrast is over. “For iodinated agents, the chemistry is done,” said Choi. “The next frontier is biology -- and the algorithms that decide how little of the agent we can safely use.”