AI-driven imaging competition heats up at KCR

AI-based reconstruction, automated measurement, and reading support expand… Competition in speed and consistency

Global equipment manufacturers and domestic AI companies present technologies centered on clinical workflows

KCR "Filters out unnecessary rescans"… Need raised for separate reimbursement funds for AI-driven imaging services

Competition in artificial intelligence (AI) for diagnostic imaging is expanding beyond lesion detection and interpretation assistance to encompass the entire imaging workflow.

As AI becomes involved in everything from image acquisition and reconstruction to automated measurements and report generation, companies are increasingly competing not simply on diagnostic accuracy but on their ability to improve examination speed and consistency while reducing clinicians’ workload.

This shift was on full display at the Korean Congress of Radiology 2026 (KCR 2026), held Sept. 9–12 at KINTEX Exhibition Center 2 in Goyang.

KCR 2026 is being held from Sept. 9 to 12.

AI incorporated into imaging equipment…automating the entire examination process

The theme of this year’s congress is “Humanity Through Imaging.” Participating companies emphasized the clinical value that healthcare professionals and patients can gain from technologies in actual examination settings, rather than focusing solely on technological performance.

While last year’s KCR centered on the hardware specifications needed to support high-performance AI and the possibility of upgrading existing equipment, a new area of competition has emerged this year -- how AI can be integrated into the imaging workflow.

Competition among global diagnostic imaging companies has shifted from obtaining clearer images alone toward simultaneously reducing the time required for image acquisition, reconstruction and measurement. AI, meanwhile, is moving beyond standalone interpretation software and becoming embedded in CT, MR and ultrasound systems themselves.

In CT and MR, the integration of hardware with AI-based reconstruction technology was particularly prominent. Siemens Healthineers applied its high-quality Quantum Iterative Reconstruction (QIR) technology to the NAEOTOM Alpha photon-counting CT system, while Philips showcased the second-generation Spectral CT 7500, which acquires conventional CT images and spectral data simultaneously, as well as the 3.0T MR BlueSeal Horizon, which is yet to be approved in Korea.

GE HealthCare, meanwhile, presented Sonic DL, which accelerates MR image acquisition, and AIR Recon DL, which reduces image noise.

(Clockwise from top left) Siemens, Canon, GE HealthCare and Philips booths

In ultrasound, where examinations depend heavily on operator technique, automated recognition and measurement technologies were a common feature.

Philips’ EPIQ Elite VM14 highlighted technology that automatically measures the kidneys and spleen and selects optimal image frames. GE HealthCare’s LOGIQ series similarly uses AI to identify and measure structures including the common bile duct, kidneys and abdominal aorta, reducing repetitive steps of testing for healthcare professionals.

Samsung Medison presented its Liver Total Solution centered on the premium R20 ultrasound system, which offers combined assessment of fatty liver, liver fibrosis and viscosity-related properties of liver tissue. Its EzSWI liver stiffness measurement technology uses AI to automatically designate the region of interest.A study by a research team led by Professor Jeong-min Lee of Seoul National University Hospital found that the technology reduced examination time by approximately 55% without a statistically significant difference in diagnostic accuracy compared with the conventional method.

Ultimately, competition in imaging equipment is shifting beyond how clearly images can be produced toward how quickly and consistently images of equivalent quality can be acquired.

Young-beom Cho, Director of Domestic Sales Team at Samsung Medison, said, “Advances in imaging technology should go beyond providing more functions and ultimately help healthcare professionals make more reliable decisions while enabling patients to undergo examinations more efficiently.”

Medical AI expands beyond detection…supports follow-up and reporting

Korean medical AI companies placed greater emphasis on tasks performed after image acquisition. Their technologies are expanding beyond diagnostic assistance that flags lesions to encompass comparison with previous examinations, follow-up management, review of interpretation results and report generation.

Lunit departed from the conventional approach of focusing on brochures and product presentations, instead setting up an interactive booth where visitors could experience the AI-assisted interpretation process firsthand.

Through the Lunit INSIGHT Challenge, visitors could interpret chest X-rays and mammograms themselves and then compare their findings with the AI results. At the product demonstration area, the company showcased the interpretation workflows of Lunit INSIGHT CXR4, Lunit INSIGHT MMG and Lunit INSIGHT DBT, its AI solution for 3D mammography interpretation.

“Sung-ho Back, a Lunit manager, explained, “AI solutions are much easier and faster to understand by experiencing them firsthand than by simply hearing an explanation. We designed the booth so visitors could interpret images themselves, compare their findings with the AI results and experience firsthand the role AI can play in an actual interpretation environment.”

(Clockwise from top left) Samsung Medison, Lunit, DEEPNOID and Coreline Soft booths

Coreline Soft presented chest CT-based medical imaging AI solutions and clinical and screening applications under the theme “AI in Radiology: Beyond Detection.”

Its AVIEW LCS Plus integrated lung cancer screening solution supports comparison with previous images and longitudinal follow-up in addition to lung nodule detection and quantitative analysis. The system enables clinicians to assess nodule size and changes according to consistent criteria across repeated examinations and link the findings to subsequent examinations and care.

DEEPNOID, meanwhile, incorporated generative AI into the interpretation process. Eight abstracts in the chest and brain imaging fields being presented by the company at KCR 2026 evaluated how accuracy, efficiency and consistency change when AI is integrated into clinicians’ interpretation workflows.

In chest imaging, the research covered a double-reading approach in which AI identifies cases that may have been missed after a physician’s initial interpretation and recommends them for review, as well as the use of generative AI to draft chest X-ray reports.

While Coreline Soft has expanded its clinical reach into post-screening follow-up management and Lunit has focused on direct comparisons between clinicians’ interpretations and AI results, Deepnoid has extended AI into the selection of cases for re-review and report generation.

Reducing only unnecessary repeat scans…will start real-time management from November

Meanwhile, the Korean Society of Radiology agreed with the need for stronger management of CT and MRI imaging histories but stressed that not all repeat imaging should be regarded as unnecessary duplicate testing.

Beginning in November, the Health Insurance Review & Assessment Service (HIRA) will pilot a system that manages patients’ CT and MRI imaging histories in real time through its medical reimbursement benefits history verification system. Once the program is implemented, medical institutions will be required to review a patient’s imaging history from the previous year before performing a scan and submit relevant information to HIRA immediately afterward.

KSR said repeat imaging may be medically necessary when performed to monitor changes in a patient’s condition or treatment outcomes, or when additional scans are needed because previous images were inadequate in quality or did not sufficiently cover the required anatomical area. By contrast, it said examinations repeated without medical justification despite usable existing images should be reduced.

Joon-il Choi, Director of the KSR’s Policy and Advocacy Institute (KSR-PAI) and professor of radiology at Seoul St. Mary’s Hospital,said, “Not every repeat CT or MRI examination should be regarded as unnecessary duplicate imaging. The policy should focus not on indiscriminately reducing the number of repeat examinations, but on accurately identifying and reducing repeat scans that lack medical justification.”

At a press briefing, the Korean Society of Radiology emphasized the importance of accurate screening rather than indiscriminately reducing the number of repeat examinations.

The society proposed measures including entering reason codes for repeat imaging, facilitating the exchange of imaging information among medical institutions, establishing objective image-quality assessments and creating reimbursement for storing outside images in picture archiving and communication systems (PACS).

On reimbursement for medical AI, the society also offered a pragmatic assessment that the National Health Insurance budget alone would be unlikely to provide compensation at the level expected by the industry.

Given Korea’s relatively low medical fees, healthcare institutions have difficulty translating productivity gains from AI adoption into additional revenue. With the government also pursuing reductions in reimbursement for diagnostic imaging, the society said it would be difficult for the National Health Insurance system to provide additional reimbursement for AI usage fees.

Choi said, “We need to fundamentally reconsider why the cost of medical AI should be borne by National Health Insurance,” and suggested that if the objective is largely to foster the industry and support exports, the government could also consider using separate funding from the industry or science and technology sectors.