Plaque Quantification: De-Risking Cardiac AI for Investors

Cardiovascular AI is moving fast, thanks to better imaging and computing, but for those of us actually running coronary CT programs, the real work is figuring out how to validate these new tools. The whole field is wrestling with evidence standards, especially for coronary plaque quantification. You have to understand the two main validation philosophies, structural plaque analysis and functional flow assessment, because they are the two pillars of modern coronary CT.

The CERTAIN Trial and Structural Plaque Analysis

Cleerly built its whole coronary CT angiography (CCTA) program on the back of the multicenter CERTAIN trial. That study was a big deal because it showed that using AI to quantify plaque actually changed how doctors treated patients, forcing them to focus on plaque burden itself. So Cleerly’s approach, which got FDA 510(k) clearance, is all about quantifying that plaque burden and stenosis, a purely structural assessment that lines up neatly with the American College of Cardiology (ACC)’s appropriate-use criteria. They’ve since built a business on it, with outcome studies showing the link between the plaque burden they measure and actual cardiac events CERTAIN trial results on plaque analysis.

Structural vs. Functional Assessment in Coronary CT

In cardiac imaging, you generally have two camps: people analyzing the structure of the plaque and people assessing the functional impact on blood flow. Cleerly is firmly in the structural camp, focused on plaque morphology and quantification. The classic example of the functional approach is HeartFlow. Their technology, centered on fractional flow reserve derived from CT (FFRct), gives you a non-invasive way to see if a stenosis is actually hemodynamically significant, something you can’t tell from anatomy alone. It’s a different piece of the puzzle, giving you blood flow dynamics to go along with the structural details from a CCTA. HeartFlow’s FFRct was one of the first AI tools for coronary physiology, and having both AI-driven structural and functional tools is what’s really pushing us forward in diagnosing and managing coronary artery disease.

Expanding Evidence Standards: Echocardiography and Digital Pathology

These tough evidence standards from coronary CT aren’t staying in their lane. They’re pushing other fields to raise their game. Look at Ultromics, they took a similar AI-quantification approach but applied it to a different modality by getting their EchoGo platform cleared for echocardiography analysis. It’s a good example of how these algorithms can bring needed quantitation to echo interpretation. But the real wake-up call came from outside cardiology altogether. When Paige AI got the first-ever FDA approval for an AI in digital pathology for primary diagnosis, it changed the conversation for everyone FDA approval for Paige AI in digital pathology. If an AI can be trusted for primary cancer diagnosis, what does that mean for the evidence we need for our cardiac tools? It means the bar for clinical validation and regulatory sign-off is only going to get higher across all of medicine.

The Shared Evaluation Base for Cardiovascular AI Innovation

So what’s the common ground for all this AI development? It’s really CCTA and cardiac risk prediction models. That’s the sandbox everyone is playing in. And it’s a big sandbox, the cardiac AI market was pegged at $2.71 billion for 2026 and is expected to hit $6.94 billion by 2030, which is a wild 26.5% compound annual growth rate. That kind of money explains the explosion of tools we’re seeing. The main regulatory route for all this activity is the FDA 510(k) pathway. Just look at the numbers: as of September 2026, the FDA had authorized over 1,600 AI-enabled medical devices, with 295 getting cleared in 2025 alone. Cardiovascular AI is the second-biggest slice of that pie, with 209 devices cleared between 1995 and 2024. A full 96% of them went through the 510(k) process, which tells you there’s a clear (if predicate-dependent) path to market FDA AI/ML medical device clearances. All this work, from structural plaque tools to functional flow analysis, is happening because we have these established evidence designs and a regulatory process that, for better or worse, can handle them. That combination is what’s building the future of cardiovascular AI.