Magnetocardiography (MCG)
A Practical Guide for Clinicians and Teams
Magnetocardiography (MCG) is a non-invasive way to measure the heart's magnetic fields, giving clinicians and cardiac teams another signal path beyond traditional ECG/EKG. The reason MCG is showing up more often right now is simple: better sensors and better computation have made it more feasible to capture faint cardiac magnetic signals and turn them into usable outputs.
SandboxAQ's AQMed focuses on applying advanced sensing and computation to clinical workflows, including MCG.
What magnetocardiography measures
The heart's electrical activity produces tiny magnetic fields. Magnetocardiography measures those fields and uses software to process and interpret them. The clinical value is not the raw signal — it is what you can consistently infer from it.
A simple way to hold it in your head:
- ECG/EKG measures electrical potential at the skin
- MCG measures magnetic fields associated with that activity
Different sensor type, different noise profile, different constraints.
Why MCG can be useful alongside ECG/EKG
Most teams evaluate MCG because it can complement existing cardiac assessment, especially when they care about:
- capturing cardiac activity with a different signal modality
- improving signal quality in challenging scenarios
- creating additional features for decision support when combined with computation
MCG is not a default replacement for existing approaches. It is a "does this add value in our workflow" question.
What makes magnetocardiography hard in practice
MCG is fundamentally a signal-to-noise problem. You are trying to measure very small magnetic signals in real environments. Common practical hurdles include:
- environmental electromagnetic noise
- motion artifacts and placement variability
- workflow friction if scanning is not operationally simple
- interpretation burden if outputs are not clinician-friendly
This is why a modern MCG system has to be designed end-to-end: sensing, shielding, capture workflow, signal processing, and interpretation.
Where AI shows up in magnetocardiography
When people say "AI cardiac" in the context of MCG, the useful interpretation is not "AI replaces clinicians." It is that computation can improve the reliability and usability of the signal. AI tends to matter for:
- denoising and artifact suppression
- feature extraction that is consistent across captures
- confidence estimates so outputs are not treated as absolute
- mapping patterns to clinically relevant categories, when validated
The underlying principle — that quantitative, physics-aware modeling improves signal extraction from complex systems — is central to SandboxAQ's Large Quantitative Models work.
What to look for in a magnetocardiography system
If you are evaluating MCG solutions, focus on operational reality.
Workflow fit
- How long does a scan take?
- How repeatable is placement and capture?
- Does it integrate into existing clinic flow?
Signal stability
- How does it perform outside ideal settings?
- What controls exist for noise and artifacts?
Interpretation outputs
- Are results presented in a way clinicians can act on?
- Are confidence or quality indicators included?
Deployment readiness
- What does maintenance look like?
- What training is required for consistent capture?
Magnetocardiography FAQ
What is magnetocardiography (MCG)?
A non-invasive technique that measures the tiny magnetic fields produced by the heart's electrical activity.
How is MCG different from ECG/EKG?
ECG/EKG measures electrical potentials at the skin. MCG measures magnetic fields associated with that same cardiac electrical activity, using a different sensing approach.
What does an MCG scan capture?
Weak magnetic signals generated by cardiac activity, which software then processes into interpretable outputs.
Why would a clinic consider MCG?
To add another signal modality alongside ECG/EKG, especially when complementary information or software-assisted interpretation could improve confidence in a clinical decision.
What are the main challenges with MCG?
Low signal strength, environmental electromagnetic noise, motion artifacts, and ensuring outputs are presented in a clinician-friendly way.
Does MCG require AI?
Not always, but AI can help improve usability by denoising signals, removing artifacts, extracting consistent features, and providing confidence estimates.
Where does interpretation come from if the signal is faint?
Typically from signal processing combined with models that translate patterns into usable outputs. SandboxAQ's Large Quantitative Models approach addresses exactly this kind of model-driven inference challenge.
Where can I learn more about SandboxAQ's MCG work?
Start with AQMed.
To explore SandboxAQ's approach to MCG and clinical sensing: