FDA Clears AI Platform That Reads Cardiovascular Risk From Routine CT Scans

HeartLung Corp. has received FDA clearance for AI-CVD, an artificial intelligence platform that automatically extracts cardiovascular and systemic health measurements from routine chest and abdominal CT scans. The clearance marks a significant step toward opportunistic screening: using scans that patients already receive for other reasons to detect cardiovascular disease risk.

What AI-CVD does

The platform analyzes existing CT images to quantify markers of cardiovascular risk, including coronary artery calcium, aortic measurements, body composition metrics, and other systemic indicators. The critical distinction is that it works on scans already performed for other clinical indications. No additional imaging, no extra radiation dose, no contrast injection, and no workflow disruption for the radiologist.

According to HeartLung, AI-CVD can be applied to approximately 40 million CT scans performed annually in the United States, which represents roughly half of all CT scans in the country. The potential to extract cardiovascular risk data from scans that were ordered for lung screening, abdominal pain, or trauma evaluation is enormous.

Why opportunistic screening matters

Traditional cardiovascular screening requires dedicated imaging: a cardiac CT for calcium scoring, a coronary CTA for stenosis evaluation, or a stress test. Each requires a separate order, a separate appointment, and often separate authorization. Many at-risk patients never receive these dedicated screenings because the referral pathway is long and the indications are not always clear-cut.

Opportunistic screening bypasses this entirely. The patient already had a CT scan for a different reason. The AI runs in the background, analyzes the images that already exist, and flags cardiovascular risk that would otherwise go undetected until symptoms appear.

Studies have consistently shown that coronary artery calcium detected incidentally on non-gated chest CTs is a strong predictor of future cardiac events. The challenge was always automation: manually measuring calcium on routine CTs is time-consuming and inconsistent. AI solves that problem.

Relevance for Indian radiology

India bears one of the highest cardiovascular disease burdens in the world. Heart disease is the leading cause of death, and the average age of first heart attack in India is nearly a decade younger than in Western countries. Despite this, dedicated cardiac screening is rare outside major urban centers.

What India does have is a growing volume of CT scans. As access to CT imaging expands through government health programs and private diagnostic chains, millions of chest and abdominal CTs are performed annually. An AI platform that extracts cardiovascular risk from these existing scans could identify at-risk patients who would never have been referred for dedicated cardiac imaging.

The infrastructure requirement is modest: the AI runs on the images post-acquisition, meaning it can be deployed as a cloud service without requiring new hardware at the scanning site. For Indian diagnostic chains operating dozens of centers, a centralized AI analysis pipeline could process scans from all locations.

Clinical integration challenges

The technical capability is clear. The clinical integration is harder. When AI flags incidental cardiovascular risk on a routine abdominal CT, who is responsible for communicating that finding to the patient? Who orders the follow-up? How does this information flow from the radiology report to the referring physician to the cardiologist?

These workflow questions are not unique to AI-CVD, but they become more urgent as opportunistic screening tools scale. Finding disease that would not otherwise have been detected is only useful if the clinical system acts on those findings. Otherwise, it creates liability without benefit.

Despite these challenges, the FDA clearance of AI-CVD represents a clear signal: the era of single-purpose imaging is ending. Every CT scan contains more diagnostic information than its original indication required, and AI is learning to extract it.