3 - Recording 12-lead Electrocardiogram with Wireless Sensing of MRI Scanner-Induced Interference

Background: The high-amplitude and variable-frequency electromagnetic interference (EMI) generated by MRI scanners represents a challenge for the high-fidelity monitoring and recording of 12-lead electrocardiogram (ECG) during MRI procedures. This issue is particularly important for interventional CMR, which requires continuous ECG monitoring as well as real-time, gradient-intensive scanning with fast imaging sequences (e.g., balanced steady-state free precession, bSSFP).
Signal-processing approaches previously developed to record ECG waveforms in the presence of EMIg used the timing and waveforms of the gradient-control signals, which were usually obtained directly from the gradient-control cabinet.^1,2 This approach limits the mobility, utilization, and scope of applications of the ECG recording systems. 
We hypothesized that the EMIg can be detected in the ECG signals using wireless sensing and can be accurately removed using real-time signal processing when the ECG system is not connected to the MRI scanner.

Methods: Pilot testing was performed in two swine using a custom 0.55T MRI scanner located at the NHLBI (modified MAGNETOM Aera, Siemens Healthineers, Erlangen, Germany), which has the same gradient system as a 1.5T scanner. The 12-lead ECG signals were recorded using an MRI-compatible ECG system with custom ECG leads (Fig. 1).³
Initial experiments were performed with the ECG recording system connected to the scanner's gradient/radiofrequency cabinet (scanner link) to obtain the gradients’ start times and waveforms. Subsequently, the link was disconnected, and the scanning was repeated using the following sequences: ungated real-time bSSFP; saturation-prepared, gated, diastolic, single-shot bSSFP; T2-prepared and navigator-gated, diastolic, 3D bSSFP; and short-TI, IR-prepared, real-time bSSFP with 3 orthogonal slices at 3, 4, 8, and 10 frames/sec.
The signal processing was implemented in the hardware as preprocessing and in the software as post-processing; it used a-priori knowledge of the ECG waveform characteristics, including the amplitudes, derivatives, and durations, to remove EMIg in real time.⁴

Results: The ECG signal quality was similar in the experiments performed with and without the scanner link. In control experiments, when the gradient sensing was disabled, the strong EMIg could not be eliminated by low-pass filtering alone (Fig. 2). By contrast, EMIg was accurately detected and removed from the signals using the wireless gradient sensing and real-time signal processing, with minimal distortion of the ECG waveforms (Fig. 3). As a result, the QRS waveforms were reliably identified and used for scanner gating during the multiple short duty-cycle CMR sequences.

Conclusion: These pilot results provide initial evidence that ECG waveforms can be reliably recorded without a scanner link and EMIg can be accurately removed using wireless sensing and signal processing in real time.