Stroke AI specialist company JLK (CEO Dongmin Kim) announced on the 10th that it presented an abstract on acute cerebral infarction detection research using the JLK AI solution at the International Stroke Conference (ISC 2025) held at the LA Convention Center in the United States from the 5th to the 7th (local time).
The abstract presentation was led by Professor Jun Sunwoo, who concurrently serves as a professor in the Department of Radiology at Seoul National University Bundang Hospital and as the Chief Medical Officer (Director) at JLK. On the evening of the 7th (local time), Professor Sunwoo presented the abstract titled "Improving Detection Sensitivity of High-Difficulty Acute Cerebral Infarction Lesions on Diffusion-Weighted Imaging (DWI) Using JLK’s AI Solution JBS-01K (JLK-DWI)" during a special session at ISC 2025.
The study focused on the fact that subtle lesions occurring in the brainstem and ultra-acute strokes on diffusion-weighted imaging (DWI), which is essential for diagnosing acute ischemic stroke, can often be missed even by experts. These lesions are classified as high-difficulty cases, challenging even for medical professionals experienced in stroke care. Professor Sunwoo suggested that such high-difficulty cases could be supplemented by JLK’s stroke solution JBS-01K (JLK-DWI).
The study was conducted on a total of 3,986 patients who underwent initial and follow-up DWI from February 2017 to November 2021. Among 250 cases, 130 were high-difficulty cases and 120 were controls.
As a result, sensitivity improved from 74.6% with medical staff diagnosis alone to 90.6% when using the JLK AI solution, and lesion segmentation accuracy (DSC) increased from 0.523 to 0.742. Furthermore, sensitivity soared to 98.0% with AI-only diagnosis, and 79.6% of cases missed by readers were additionally detected.
The five medical professionals participating in the study evaluated that diagnostic confidence for high-difficulty cases was higher with AI assistance. Based on this, the research team concluded that JLK’s AI solution JBS-01K (JLK-DWI) significantly improved diagnostic performance and lesion segmentation accuracy for subtle lesions or ultra-acute strokes.
Professor Jun Sunwoo stated, "Through this abstract presentation, we were able to widely showcase the excellent validation capability of JLK’s stroke AI solution to medical institutions worldwide," adding, "We hope the use of stroke AI solutions that alleviate the burden and fatigue faced by medical staff during diagnosis of high-difficulty cases and provide appropriate treatment and rehabilitation plans to patients will become more active."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
