Chest pain accounts for a large number of emergency department visits, and distinguishing between ischaemic heart disease (IHD) and non-cardiac chest pain is a major challenge. Currently, patients are assessed via a time-consuming process based on electrocardiography (ECG) and blood tests. But almost three-quarters of patients with chest pain do not have a cardiac related condition. The ability to rapidly rule-out IHD could improve patient care and save hospital resources.
One possible approach is magnetocardiography (MCG), which maps the magnetic fields generated by electrical activity in the heart. MCG offers improved diagnostic capability over ECG, as it can be used to help physicians identify patients not presenting with active myocardial ischaemia.
The heart’s magnetic field, however, is millions of times smaller than the background noise in a hospital. To detect such small fields, MCG systems normally use superconducting quantum interference devices (SQUIDs), which are bulky, expensive, and require liquid helium cooling with extensive shielding. To overcome these limitations, a team at the University of Leeds developed a portable magnetometer to perform MCG scans at the patient’s bedside. They went on to set up Creavo Medical Technologies to commercialize the device.
“We needed to develop something entirely new,” explained Ben Varcoe. “The key technical development was using a different quantum effect called the ‘Hanbury Brown and Twiss effect’. This is used by large astronomical microwave antenna arrays to separate distant signals from closer noise sources by looking for correlations in the data. We employ the same effect but in reverse: we use the antenna array to remove the distant background so we can see the local field from the heart.”
Varcoe and colleagues recently demonstrated that the device can rapidly distinguish patients with non-ischaemic chest pain from patients with IHD (PLOS ONE 13 e0191241).
To create the portable MCG device, the research team employed induction coil magnetometers, which offer high sensitivity, are inexpensive, do not require cooling and can be run on batteries. Such a system can detect the magnetic field of the heart with sufficient sensitivity and low inherent noise (Biomed. Phys. Eng. Express 3 015008).
The researchers used two prototype devices to perform a technical performance study and a pilot clinical study. The former included patients with suspected IHD and healthy age-matched volunteers, plus a subgroup of patients with non-ST-elevated myocardial infarction (NSTEMI); the latter was conducted in NSTEMI patients admitted for chest pain and a control group of non-IHD patients with chest pain. Additional data were collected from a young healthy reference group.
They divided the participant data into three groups. Group A included 70 IHD patients: 55 from the first study and 15 from the second. Group B included 69 controls: 51 from the performance study with no IHD and 18 from the clinical study, who had non-ischemic chest pain. Group C included 37 healthy volunteers.
MCG scans were recorded for 10 min in an unshielded room. The researchers baseline-corrected and averaged the MCG signals to increase the signal-to-noise ratio, extracted 10 MCG predictors from the data, and used logistic regression modelling to evaluate these candidate predictors. Three predictors showed promise for differentiating patients from age-matched controls: QR_peak, RS_peak and RS_MMR.
“Q, R and S are features on an ECG trace that can be used as diagnostic markers; the same features are seen in the magnetic field traces so it makes sense to use the same notation,” Varcoe explained. “In reaching a diagnosis, we use the slope of the transitions in the trace. So QR, for example, is the slope between Q and R.”
The researchers investigated three logistic regression models: Model 1 used Groups B and C as the control group; Model 2 used Group B; and Model 3 used Group C. The patient group (A) was the same across all three. They examined the ability of the models to distinguish patients and controls using the area under the receiver operator characteristics curve (AUC). The ability to rule-out subjects was assessed via sensitivity, specificity and negative predictive value (NPV).
All models showed respectable rule-out ability. Model 1 yielded an AUC of 0.82, specificity of 33.0%, sensitivity of 98.6%, and NPV of 99.3%. Model 3 achieved a near perfect separation between groups, with an AUC of 0.96 and specificity of 78.4% (sensitivity and NPV both 100%). Separation between the groups was lower in Model 2, with an AUC of 0.75, specificity of 20.3%, sensitivity of 94.3% and NPV of 95.2%. Cross-validation revealed that, using Model 1, the magnetometer ruled-out 35.0% of the control group with 97.7% NPV.
More recently, the team at Creavo has produced prototype clinical devices that are CE marked and FDA approved for clinical investigations. “The current device developed by Creavo is a substantial improvement over the original devices that we created at the University of Leeds,” Varcoe told Physics World. “It has 37 sensors, increased from 15, and uses a set of bespoke sensors that are smaller, lighter and more sensitive.”
Creavo is now using these new systems in a multi-centre clinical trial to evaluate performance in an emergency department. “They have been used in five hospital emergency departments in the UK by a huge number of staff to collect the magnetocardiograms of 750 patients,” Varcoe noted. “A similar trial is also about to launch in the US.”
“From here we will assemble the information that we have gathered, including user experience and patient feedback, to evolve the current design into a device that can be used to aid emergency physicians in the exclusion of active, acute myocardial ischaemia in patients presenting with symptoms consistent with chest pain of cardiac origin.”