CHICAGO, Ill. -- By Lee Han-soo/Korea Biomedical correspondent -- Coreline Soft continued to maintain its strong presence at Radiological Society of North America (RSNA) 2024 in Chicago, Illinois, marking nearly a decade of consecutive participation since 2015.

The Korean AI company has consistently drawn attention at the largest annual gathering in the field of radiology by addressing radiologists' evolving needs with innovative solutions.

As one of the leading Korean medical AI companies, Coreline Soft has established a tradition of unveiling enhanced software and new products at each year's conference, demonstrating its commitment to advancing radiological care.

In a recent interview with Korea Biomedical Review during the conference, Coreline Soft CTO Yi Jae-youn highlighted their expansion from lung cancer screening to broader clinical applications. The Korean AI medical imaging company has been making significant strides in international markets.

"While we've been concentrating on lung cancer screening, our technology actually pursues comprehensive mortality reduction by detecting various incidental findings," Yi said.

Yi noted that their current focus extends beyond the traditional "Big 3" diseases -- lung cancer, emphysema, and cardiovascular disease - to include conditions like interstitial lung abnormalities (ILA).

"Currently, there's increasing research interest in finding ILA during lung cancer screening, as these high-risk individuals are already getting regular CT scans annually or biennially," Yi said. "While we don't call these findings ILD (Interstitial Lung Disease) yet, early detection could be crucial, as ILD typically has poor outcomes once symptoms appear."

Yi noted that this focus on ILA detection aligns with broader screening objectives.

"When patients come to hospitals with symptoms of ILD, treatment options are often limited due to existing fibrosis,” he said. “The guidelines for managing ILA findings are still in early discussions, but we believe this could be the next important target after our established Big 3 diseases in chest screening."

Regarding Coreline's recent U.S. entry, Yi emphasized the importance of adapting to different clinical workflows.

"Each partnership brings unique value," Yi said. "With UMass Memorial Medical Center, we're learning that while AI performance is essential, American radiologists are particularly focused on how seamlessly the technology integrates into their existing workflow systems, whether it's with PACS or other platforms."

Yi highlighted the cultural differences in AI adoption between markets.

"In Korea, our products are primarily used in lung cancer screening programs, which made workflow adoption relatively easier because screening programs have unique workflows," he noted. "However, in the U.S., our nodule detection solutions are being integrated into general clinical practice, where physicians prefer to maintain their established workflows.”

The customization of reporting formats and visualization methods also varies significantly between markets.

"It's not just about PACS integration," Yi noted. "American users have different preferences for report formats, report content, and even how images are captured and displayed,” the CTO said. “The process of going live in the U.S. requires us to customize these elements to match their specific needs, while maintaining the core AI technology."

The company is also keeping a close eye on emerging technologies like photon-counting CT.

The company is closely monitoring the evolution of photon-counting CT technology, which has been highlighted by major manufacturers as the future of CT imaging. "We're watching the photon-counting CT developments with great interest, particularly for lung imaging," Yi said. "The enhanced resolution could potentially allow us to visualize much finer airways, especially for conditions like small airway disease, which has generated considerable interest from our research partners."

However, Yi pointed out several practical challenges in implementing AI with this new technology.

"While the image resolution is significantly better, there are practical challenges," he said. "Our current AI models are trained on conventional CT resolution.”

When applied to photon-counting CT, the processing time increases about eight-fold due to the higher resolution data and the company is actively working on optimizing this, Yi added.

The transition to photon-counting CT also presents unique challenges for AI development.

"Currently, our AI is trained to detect abnormalities at the resolution level of conventional CT and the models tend to ignore features smaller than what they're trained to recognize," Yi explained.

As a result, Yi mentioned that several European hospitals with photon-counting CT installations have approached Coreline for collaborative research.

"We're already conducting pilot studies with these institutions," he said. "However, as some pulmonary researchers have noted, we're still waiting to identify the true 'killer application' for photon-counting CT.”

While Coreline’s current solutions are functional on these systems, there are still aspects that need refinement for optimal clinical use, Yi added.

Moving forward

While Coreline Soft has established itself in chest imaging, including cardiac applications, Yi revealed plans for expansion into other critical areas of medical imaging.

"Currently, we're particularly interested in advancing into emergency medicine solutions, starting with the Korean market," Yi said. "Beyond our established chest solutions, we've developed brain hemorrhage detection for emergency settings, and we're actively preparing solutions for emergency chest pain evaluation."

Yi emphasized that the emergency room, like screening programs, is another area where AI can make a significant impact.

"ER physician burnout is enormous and we're developing solutions for brain hemorrhage detection and chest pain evaluation in emergency settings,” he said.

Yi highlighted how outdated some current emergency radiology practices are, noting instances where doctors still use FaceTime to view PACS screens remotely.

"As someone from the IT industry, seeing these practices still in use suggests there's ample opportunity for enhancement," he said. "While AI will be the first step, we're looking at comprehensive solutions for managing emergency imaging workflows."