At RSNA 2023, Lunit engaged in eight in-booth meetings with global leaders to enhance the impact of its AI solutions. During this event, KQ, a prominent provider of AI-driven cancer diagnostics and treatments, unveiled eight innovative research discoveries and expert insights.
KQ will showcase eight groundbreaking abstracts at the RSNA 2023 Annual Meeting in Chicago, scheduled from November 26 to November 30. Additional findings, comprising five presentations, may be highlighted as ePosters, with three delivered orally.
Lunit’s research delves into the effectiveness of a newly developed normal filtering (NF) AI model in autonomously interpreting normal chest radiographs during an oral presentation. The study evaluates the integration of Lunit INSIGHT CXR, a commercially available AI for detecting predefined findings, the percentage of chest radiographs suitable for exclusion from reporting, and the AI’s sensitivity in detecting abnormalities. The NF model, with a specificity threshold of 50%, demonstrated an average sensitivity of 97.8%, resulting in a 22% overall reduction in chest radiographs requiring reporting. Additionally, the commercially available AI identified 16.7% of previously undetected clinically significant defects, serving as a valuable safety net. The research underscores Lunit INSIGHT CXR’s role as a critical safety measure to prevent oversight of potential defects and highlights the potential of routine screening AI in alleviating radiologists’ workload. Lunit plans to formally introduce the NF design shortly, showcasing its dedication to cutting-edge AI solutions in radiology.
Furthermore, Lunit explores mammographic parenchymal patterns and temporal changes to develop an AI model capable of predicting future breast cancer risks. Utilizing an extensive dataset of 16,113 full-field digital mammography scans from over 9,000 individuals, the AI model exhibited enhanced risk prediction capabilities with AUCs of 0.75 at 1-year, 0.76 at 2-year, and 0.73 at 3-year intervals. This study validates the feasibility of an AI predictive model for identifying mammographic parenchymal characteristics, enhancing risk stratification through temporal changes, and expediting personalized breast cancer screening.
Insights from the ScreenTrustCAD trial’s sub-study are also shared by a Swedish research team led by Dr. Karin Dembrower, Senior Physician at Capio S:t Göran Hospital Mammography Clinic. Published in Lancet Digital Health, ScreenTrustCAD represents a groundbreaking exploration of AI in large-scale population-based breast cancer screening. The study reveals that Lunit INSIGHT MMG, along with a single reader, outperformed two human readers in cancer detection without increasing recalls.
This sub-study examined variations in consensus recall decisions based on whether radiologists or Lunit INSIGHT Firm initially classified the examination as abnormal. With AI, user 1, and readers 2, a total of 5,515 examinations were flagged for further evaluation. Of the 2,501 consensus cases not flagged by AI but by one or two radiologists, 25% were recalled, with 0.8% resulting in cancer detection, yielding a 3.4% positive predictive value (PPV). In contrast, 24% of the 3,014 consensus cases flagged by AI (and 0 to two radiologists) were recalled, with 7.2% leading to cancer detection, resulting in a 29% PPV.
Notably, the PPV for examinations flagged by AI was nearly ten times higher than those flagged solely by radiologists for the cases recalled during the consensus review. The study suggests that participants in review discussions may benefit from insights into their decision-making processes, prompting potential redesigns to maximize the utility of AI interpretations.
Brandon Suh, CEO of Lunit, emphasized, “Our analyses showcase how AI can enhance radiology, from streamlining radiologists’ tasks to predicting breast cancer risks effectively.” He further added, “I am confident that these studies significantly contribute to the new era of imaging and personalized cancer care.” Visitors are encouraged to attend Lunit’s RSNA 2023 AI Showcase booth (#4165) to explore these studies and engage in eight enlightening in-booth sessions featuring distinguished speakers such as Dr. Karin Dembrower and former IBM Watson Health Chief Medical Officer David Gruen. With the Lunit Information suite deployed in over 3,000 healthcare facilities worldwide, attendees can witness firsthand its transformative impact on the future of radiology.
Key Highlights about Lunit at RSNA 2023: About Lunit
Lunit, an adept AI firm specializing in deep learning, is dedicated to eradicating cancer. The primary objective is to develop AI-driven therapeutic and diagnostic solutions that offer each patient optimal diagnosis and care. Leveraging state-of-the-art technology, Lunit focuses on creating innovative medical image analytics and AI-based biomarkers.
Established in 2013, Lunit has garnered acclaim for its advanced technology and its application in medical imaging. The agency’s research is featured in reputable peer-reviewed publications like the Journal of Clinical Oncology and Lancet Digital Health, as well as international conferences such as ASCO and RSNA. With FDA approval and the CE Mark, the flagship Lunit INSIGHT suite is deployed in over 3,000 hospitals and healthcare facilities across 40+ countries. With a global presence and headquarters in Seoul, South Korea, Lunit continues to drive advancements in healthcare through cutting-edge AI solutions. 
