- AI detects early heart vessel changes before blood flow drops
- Helps patients with chest pain or heart disease risk
- Still in testing — not yet available in clinics
This new tool could catch heart disease earlier than ever.
You wake up with a tightness in your chest. Not sharp. Not scary. Just… off. You’ve had it before. Your doctor says your arteries aren’t blocked. So why do you still feel this way?
Now, a new study suggests we’ve been missing something — not blockages, but how well your heart’s small blood vessels work. And AI may be able to see it first.
The Hidden Problem
Millions of people have chest pain but normal-looking arteries on scans. They’re told their heart is fine. Yet many still feel unwell. Some even have heart attacks later.
This is often due to a condition called coronary microvascular dysfunction — problems in the tiny blood vessels of the heart. These vessels don’t show up well on standard tests.
It affects up to 50% of patients with chest pain and no major blockages. Women are more likely to have it than men. But until now, it’s been hard to spot early.
Doctors usually wait until blood flow drops or damage occurs. By then, it’s already advanced.
Old Clues, New Tools
For years, we focused on big blockages. If an artery was 70% closed, we knew it needed stents or bypass surgery.
But we missed the early warning signs — subtle changes in vessel walls, blood flow patterns, or tissue stress.
We had tools like CT scans to see anatomy. And SPECT scans to measure blood flow. But they didn’t talk to each other.
Now, AI is connecting the dots.
What scientists didn’t expect
AI can now analyze routine heart CT scans and find patterns invisible to the human eye.
In this study, researchers used AI to study coronary CT angiograms — common scans that show heart arteries in 3D.
The AI didn’t just look for blockages. It measured tiny changes in vessel texture, shape, and surrounding tissue — signs of early stress.
Then it compared those findings to two key markers:
- FFR-CT: a computer estimate of blood flow
- CFR: actual blood flow measured during stress tests
Here’s the catch: these AI-detected changes showed up before blood flow dropped.
Like a traffic jam before the crash
Think of your heart’s arteries like highways.
A major blockage is like a car crash — traffic stops, and everyone knows there’s a problem.
But what about the early signs? The potholes. The narrowing lanes. The backup starting miles before the crash?
That’s what AI is now spotting — the early wear and tear on the “roads” of your heart.
It’s like predicting a traffic jam by studying road conditions, not waiting for the crash report.
The study looked at 251 patients who had both a heart CT scan and a nuclear stress test (CZT-SPECT).
All had suspected or known heart disease. The AI analyzed their CT scans for subtle vessel changes.
Then researchers compared those AI findings to actual blood flow measurements.
Patients with abnormal AI scan results were much more likely to have poor blood flow — even if their arteries looked open.
The AI signals strongly matched low coronary flow reserve (CFR) — a sign the heart can’t get enough blood during stress.
It also lined up with FFR-CT, another measure of reduced flow.
In plain terms: the AI could flag at-risk hearts before traditional tests showed anything wrong.
One expert noted: “We’re moving from anatomy to function — and doing it earlier.”
This doesn’t mean this treatment is available yet.
But there’s a catch.
The study was retrospective — meaning it looked back at existing data.
It wasn’t designed to prove AI improves outcomes. Just that it can detect early signs linked to real problems.
Also, the AI was trained on a specific type of CT scan. It may not work the same on older machines or different hospitals.
And while it found strong links, we still need proof that acting on these AI signals leads to better health.
Heart disease is still the #1 killer worldwide.
Many people are diagnosed too late — after damage is done.
If we can catch problems earlier, we could prevent heart attacks, reduce hospital visits, and tailor treatments sooner.
Imagine getting a routine CT scan and learning your heart vessels are under stress — even if they’re not blocked.
That could mean earlier lifestyle changes, better meds, or closer monitoring.
Right now, this AI tool is still in research labs.
It’s not part of standard care. You can’t ask for it at your local hospital — yet.
But if you have chest pain and normal artery results, talk to your doctor about microvascular disease.
Ask if advanced blood flow testing (like CFR or FFR-CT) might help.
And stay tuned — tools like this could become routine in the next 5–10 years.
The study only included patients who already had scans — mostly those with symptoms.
It wasn’t a diverse population, so results may not apply to everyone.
And since it used older data, we don’t know how well AI predictions hold up over time.
Researchers are now testing this AI in larger, more diverse groups.
Next steps include real-time trials — where AI guides treatment decisions and tracks patient outcomes.
If proven effective, it could be built into standard CT scan software, giving doctors an early warning system without extra tests or radiation.
It’s not a magic fix. But it’s a powerful step toward smarter, earlier heart care.