How do you scan a heart while it is working? An MRI-and-exercise combo shows one way

Inside a 1.5-tesla MRI scanner at University Hospital Düsseldorf, a healthy volunteer lies on their back, feet clipped into a supine bike, while a real-time cardiac MRI records moving images of their beating heart and a face mask logs each breath. The imaging half is a heart scan that uses magnetic fields to capture a beating organ in motion rather than at a single frozen moment. The breathing half is cardiopulmonary exercise testing, the same mask-and-pedal protocol used in stress labs to measure how the heart, lungs, and muscles handle effort. For the first time in a peer-reviewed dataset, those two streams are being merged into one timeline.

Each stream is decades old. The Düsseldorf team's contribution is to run them at the same time inside the same magnet and align the resulting numbers to a single heartbeat-by-heartbeat record.

That fusion matters because today's two clinical tools answer different halves of the same question. A standard exercise MRI, established for cardiac imaging during effort by La Gerche and colleagues in 2017, can show whether the heart is pumping well under stress but cannot measure how much oxygen the muscles are pulling from the blood. A cardiopulmonary exercise test can measure that pull, through breath-by-breath gas exchange, but cannot see the heart that is supposed to be delivering the oxygen. Patients whose exercise capacity is reduced for unclear reasons often finish both tests still unexplained.

The Röwer et al. study, published 21 July 2026 in Frontiers in Cardiovascular Medicine and indexed on PubMed as PMID 42643259, tested the merged protocol on 12 healthy adult volunteers. The protocol stacked conventional cine MRI, real-time MRI, phase-contrast flow imaging of the aorta, and an MR-compatible spirometer (Geratherm Respiratory) coupled to a COSMED silicone face mask, with ECG and respiratory signals used to bin images into a synthetic heartbeat. Volunteers pedaled at a submaximal effort while supine.

The headline numbers, drawn from the paper: indexed oxygen consumption rose from 201 to 309 mL/min per square meter of body surface as exercise ramped up. Cardiac index, the heart's output per minute adjusted for body size, rose from 3.38 to 4.24 liters per minute per square meter. Left ventricular volumes dipped slightly. Ejection fraction, the share of blood the left ventricle pushes out with each beat, did not change. The Fick-derived arteriovenous oxygen difference, a calculated estimate of how much oxygen the muscles are pulling out of each unit of blood, rose from 5.9 to 7.3 milliliters per deciliter. The authors read that rise as evidence the merged protocol is picking up a peripheral, muscle-side, response to exercise in addition to the heart's central response.

This is the discrimination the technique is built to make. A patient who hits their exercise ceiling with a healthy-looking heart but a flat arteriovenous oxygen difference is limited peripherally, by muscles that cannot extract what the blood offers. A patient with a failing heart and a robust peripheral extraction is limited centrally. Neither today's resting scans nor either test in isolation sorts those two apart.

Two honest limits travel with the result. The first is scale. Twelve healthy adults at a single center, on a single 1.5-tesla scanner, exercising only to a submaximal level, only while lying on their backs, is a feasibility dataset, not a clinical one. The authors frame the work explicitly as groundwork for future studies in heart failure and impaired cardiac reserve, not as a new clinical test. The second limit is a methodological one. The arteriovenous oxygen difference is computed from the Fick equation using an assumed arterial oxygen content, because the scanner does not sample arterial or mixed-venous blood invasively. The mechanistic central-versus-peripheral claim is therefore method-grounded, not yet outcome-validated. Real patient studies will have to clear that bar, and to compare the merged readout against what conventional stress imaging and cardiopulmonary testing already produce together.

The next move is a patient cohort, ideally one with suspected heart failure with preserved ejection fraction, a population in which the central-versus-peripheral question is a daily clinical puzzle. The Düsseldorf team says the dataset is the basis for reference values and for that future work. Whether the merged protocol can hold up in patients with stiff hearts, leaky valves, or blocked coronary arteries is the question that will decide whether the tool ever leaves the research magnet.