HeartLung unveils AI upgrade to coronary calcium scoring

HeartLung Corporation says a new peer-reviewed study introduces Agatston-2.0, an AI-based update to the 36-year-old coronary artery calcium score. In a pooled analysis of 3,965 people with a traditional score of zero, the method found hidden calcium in 21.7% and identified a higher long-term heart disease risk.

Why it matters: - Agatston-2.0 is designed to update a core preventive cardiology tool that has guided risk assessment for decades. - The method aims to reduce false reassurance from a traditional CAC score of zero by finding subthreshold calcium that conventional scoring can miss. - Better risk discrimination could affect prevention decisions, follow-up timing and treatment discussions for patients labeled low risk.

What happened: - HeartLung Corporation announced the publication of a peer-reviewed study introducing Agatston-2.0, a next-generation AI-based coronary artery calcium scoring method. - The study was published in the American Journal of Preventive Cardiology under the title “Agatston-2.0: A Next-Generation AI-Based Coronary Calcium Quantification Approach to Improve Risk Stratification Among Individuals with Zero Agatston Scores – Part I.” - The work was led by Morteza Naghavi, MD, founder and president of HeartLung Corporation, and Arthur S. Agatston, MD, the original developer of the Agatston score. - The study was released June 12, 2026.

The details: - The original Agatston score, introduced in 1990, uses a fixed 130-Hounsfield-unit threshold and conventional 2.5- to 3-mm CT slice thickness. - Agatston-2.0 uses AI-based coronary segmentation and continuous voxel-wise calcium quantification instead of rigid thresholds. - The method is intended to detect very small, low-density, fragmented or partially calcified plaques that traditional scoring can overlook. - Researchers pooled 3,965 participants with CAC=0 from the Multi-Ethnic Study of Atherosclerosis and the Framingham Heart Study. - Agatston-2.0 detected AI-derived coronary calcium in 862 participants, or 21.7% of those classified as CAC=0 by conventional scoring. - Over 20 years, coronary heart disease incidence was 7.7% in participants with AI-CAC greater than zero, compared with 3.8% in those with AI-CAC of zero. - After adjustment for traditional cardiovascular risk factors, AI-CAC greater than zero remained independently associated with incident coronary heart disease. - Agatston-2.0 also predicted later conversion from CAC=0 to a positive traditional CAC score on repeat CT imaging. - The study included investigators and collaborators from HeartLung.AI, Cornell University, The Agatston Center, UCLA / Harbor-UCLA and The Lundquist Institute, University of California Irvine, Icahn School of Medicine at Mount Sinai, Stanford University, Houston Methodist, Cedars-Sinai, Kaiser Permanente, University Medical Center Groningen, University Hospital Basel and other institutions. - The publication lists DOI 10.1016/j.ajpc.2026.101698.

Between the lines: - The findings try to refine the “Power of Zero” without discarding the traditional CAC score’s clinical value. - The study’s main argument is that some patients with a zero score may still have early calcified disease below standard detection thresholds. - That distinction could matter most for people near treatment or follow-up decisions, where a slightly higher-risk subgroup changes management. - The release positions Agatston-2.0 as an evolution of the most established CT biomarker in preventive cardiology, not a replacement for all prior risk tools.

What's next: - The authors say Agatston-2.0 may help personalize cardiovascular prevention by separating higher-risk patients within the CAC=0 group from an even lower-risk AI-CAC=0 group. - HeartLung says the method could become a new standard if additional cohorts validate the findings and the system is implemented at scale. - The company also describes AutoCAC as an automated Agatston-1.0 product and AI-CAC as its more sensitive Agatston-2.0 platform.

The bottom line: - HeartLung is betting that AI can make coronary calcium scoring more sensitive without abandoning the familiar framework that preventive cardiology already uses.