Skip to content

Human Cardiac Measurements with Diamond Magnetometers

FieldValue
JournalOpen MIND
Published2026-01-26
AuthorsMuhib Omar, Magnus Benke, Shaowen Zhang, Jixing Zhang, Michael Kuebler
DOI10.48550/arxiv.2601.18843
Citations1

Three independently developed NV-diamond magnetometers detected human cardiac magnetic signals without touching the chest after averaging 300 to 12,000 heartbeats. The 12,000-heartbeat trace came from a zero-field sensor with 13 pT/√Hz noise density in a multilayer shielded room. Reported magnetic noise densities ranged from 7 to 26 pT/√Hz.

The three groups used optically detected magnetic resonance, or ODMR, for sensing at room temperature. JGU operated without a static bias field and recovered sensitivity with an oscillating magnetic field. ZAQuant and Q.ANT used a 1 mT bias and microwave amplitude modulation.

GroupDiamond dimensionsSensitivityMagnetic biasEnvironment
JGU0.5 x 0.5 x 0.18 mm13 pT/√HzZero DCMultilayer shielded
ZAQuant0.5 x 0.5 x 0.5 mm7 pT/√Hz1 mTSingle-layer shielded
Q.ANT1 x 1 x 0.5 mm26 pT/√Hz1 mTUnshielded

The paper reports active sensing volumes below 0.5 mm3. This small volume supports differential sensing for common-mode noise rejection.

JGU applied a 6.12 kHz magnetic modulation with a µT amplitude along [100]. It addressed NV hyperfine levels with 2.87 GHz microwaves mixed with a 2.16 MHz radio-frequency signal. ZAQuant aligned its 1 mT field with one NV axis, while Q.ANT used a Halbach ring to resolve all four orientation pairs.

JGU used a truncated pyramid with a 0.18 mm height and a 0.5 x 0.5 mm base. The 12 mm sensor head sat approximately 1 cm from a seated subject’s chest. Data were recorded in 22 min intervals.

Electrodes on the subject’s back supplied an ECG trigger for averaging across heartbeats and accounting for heart-rate variability. The diamond sensor itself remained non-contact. The ECG electrodes supplied only the averaging trigger.

No MCG trace appeared in the NV data in real time. The displayed signal required 12,000 averages. The 13 pT/√Hz JGU value was averaged between 20 Hz and 30 Hz.

ZAQuant recorded a seated subject approximately 1 cm away and applied narrow band-stop filters around 50 Hz, 60 Hz, and harmonics. After filtering, 300 averages exposed the MCG trace.

Q.ANT placed a subject directly below its sensor in an unshielded environment. Four notch filters around 8.8, 14.7, 16.7, and 50 Hz preceded a 30 Hz Butterworth low-pass filter and an 8 Hz high-pass filter. Each notch filter used a quality factor of 30. The Q.ANT trace appeared after 2,000 averages.

Sensitivity analysis used 9 s JGU traces, 60 s ZAQuant traces, and 10 s Q.ANT traces. The JGU analysis used a Hann window before Fourier transformation. The amplitude spectrum was divided by the square root of the effective noise bandwidth. Residual laser noise set the JGU magnetically insensitive noise floor.

MCG measures magnetic fields from cardiac currents and is less affected by tissue conductivity than ECG. Recording at two chest positions with opposite MCG peak polarity ruled out electrical cross-talk from the diamond electronics.

To map the cardiac field, the authors used a 47-sensor FieldLine OPM array in a two-layer mu-metal shield. The array sensors had 2.5 cm spacing and at least 5 cm chest stand-off. OPM MCG signals appeared in real time before averaging.

During roughly 5 min of acquisition, a 3 to 30 Hz band-pass filter and 100-heartbeat averaging produced the OPM traces. Diamond measurements then used positions A and B with 12,000 and 6,000 heartbeat averages.

QRS complex amplitudes were comparable between the OPM and diamond measurements, and both agreed with literature values. The T-wave shapes differed markedly. Position, stand-off distance, sensitive volume, and directional variation across the sensing volume can change T-wave morphology. This comparison supports the cardiac origin and magnetic character of the averaged signal.

Uniform environmental noise can exceed cardiac fields by several orders of magnitude. ZAQuant tested two identical NV sensor heads separated by 78 mm inside background-field Helmholtz coils. The coils were about 1.5 m in diameter and sat in a single-layer shielded room. Applied test fields had magnitudes of several hundred nanotesla.

Simple subtraction cancelled the test field and substantially reduced total noise in the differential trace. The individual sensors showed noise densities above 100 nT/√Hz at low millihertz frequencies. The differential trace was suppressed by almost two orders of magnitude.

This measured result remains below the cited benchmark for a well-matched gradiometer, which can exceed 60 dB, or a factor of 1000. The ZAQuant linear range degraded beyond about ±500 nT because of the sensor’s roughly 200 kHz magnetic resonance linewidth.

The authors state that active frequency tracking and closed-loop feedback could extend operation to field variations of several microtesla. That extension was proposed rather than measured during the gradiometer test. The test used controlled fields rather than live human MCG, so it established noise rejection rather than clinical detection.

The current result depends on averaging and does not establish live detection. For clinical MCG and MEG, the authors identify sub-pT/√Hz sensitivity as the relevant range. SQUID and OPM systems cited by the paper reach this range.

The authors’ estimate uses an average 25 pT R-peak amplitude, an approximately 30 Hz bandwidth, and a 50 pT single-shot RMS noise floor. For an SNR of 5, the required sensitivity is about 1 pT/√Hz. That target is roughly one order of magnitude below the sensors used here.

The continuous-wave protocols are approaching their practical sensitivity limits. Photon shot noise falls as 1 divided by the square root of detected fluorescence power. More optical power causes power broadening, sample heating, and laser-induced decoherence.

Infrared absorption, pulsed magnetometry, cavity-enhanced readout, and many-body methods are proposed routes toward lower noise. Pulsed magnetometry has reached 490 fT/√Hz in another NV setup, according to the cited work.

Flux concentrators offer a more immediate option and can raise the local signal by more than two orders of magnitude. They reduce spatial resolution, which the authors consider acceptable for cardiac monitoring at centimeter scale. MEG has a stricter requirement below 10 fT/√Hz, so the same gain would need gradiometric suppression above 60 dB.

The experiments used non-invasive self-experiments and did not test diagnostic accuracy in patients. The unshielded Q.ANT result still required 2,000 averages, and no real-time MCG trace was reported. Gradiometry was validated under applied fields, not through an unshielded live cardiac recording. The T-wave comparison also shows that position and sensitive volume affect waveform shape.

For device engineering, the first choice is between zero-field and bias-field operation. Zero-field control gave stable long-term resonance locking, while bias-field sensors support scaling toward unshielded bedside use. Material geometry must preserve the small active volume needed for gradiometric common-mode rejection.

JGU mounted the sensing diamond and lens on a diamond plate for thermal management inside an aluminum housing. The Q.ANT element contained about 1.7 ppm NV centers, measured by EPR spectroscopy. Optical collection, differential photodetection, and thermal management remain part of the sensor head.

A device team must set sensitivity, stand-off, and averaging against the intended clinical workflow. The paper supports diamond magnetometry as a route to contactless MCG, but its clearest trace still required 12,000 averaged heartbeats.

The MCG devices need NV-center single-crystal diamond sensing elements with defined axes, optical access, and submillimetre geometries. 6CCVD’s diamond materials catalog lists MPCVD-grown single-crystal plates from 0.1 to 500 micrometers thick, with polishing to Ra below 1 nm. For this MCG application, the engineering support page provides the custom fabrication route.

View Original Abstract

We demonstrate direct, non-invasive and non-contact detection of human cardiac magnetic signals using quantum sensors based on nitrogen-vacancy (NV) centers in diamond. Three configurations were employed recording magnetocardiography (MCG) signals in various shielded and unshielded environments. The signals were averaged over a few hundreds up to several thousands of heart beats to detect the MCG traces. The compact room-temperature NV sensors exhibit sensitivities of 6-26 pT/Hz^(1/2) with active sensing volumes below 0.5 mm^3, defining the performance level of the demonstrated MCG measurements. While the present signals are obtained by averaging, this performance already indicates a clear path toward single-shot MCG sensing. To move beyond shielded environments toward practical clinical use, strong noise suppression is required. To this end, we implement NV-based gradiometry and achieve efficient common-mode noise rejection, enabled by the intrinsically small sensing volume of NV sensors. Together, these multi-platform results obtained across diverse magnetic environments provide a solid foundation for translating quantum sensors into human medical diagnostics such as MCG and magnetoencephalography (MEG).