BS19 Diagnostic 12-lead ECG recordings in the 3-Tesla CMR bore at rest and during adenosine stress perfusion- results from MyoFit46

Background Current electrocardiographic (ECG) devices built into cardiovascular magnetic resonance (CMR) scanners have a narrow bandwidth (0.5–60Hz) and signals suffer from distortion due to magnetic field gradient artefact and magnetohydrodynamic effects, so ECGs are non-diagnostic except for R-peak detection. Despite advanced QRS detection algorithms, incorrect R peak detection hampers image acquisition, especially at higher field strengths, thus prolonging scan time and producing suboptimal images. Previous feasibility studies at 1.5 and 3 Tesla (T) using external ECG devices were small: n=14 and n=4 healthy volunteers respectively. We implemented a novel hardware and software system for integrating an external device into a clinical 3T CMR scanner to derive diagnostic 12-lead ECGs from inside the scanner bore and during stress perfusion acquisition. Methods Standard 12-lead ECGs were first recorded on participants of the MyoFit46 cardiovascular sub-study of the National Survey of Health and Development outside the CMR environment using conventional electrode placement. CMR was then performed using Siemens Prisma 3T magnet. Three electrodes (each with 3 measurement and 1 reference electrodes, figure 1a) were applied to the chest of participants. Short leads connected each electrode to a sensor for signal amplification and digitalisation (figure 1b). Signals were then transmitted via 10 m-long dual fibre-optic cables running through a penetration hole into the adjacent room to reach an electronics signal module and connected laptop (32 cores, 5.2GHz) equipped with easyG/truzyG/Epsidy software (figure 1c). ECGs were recorded for 30 seconds prior to image acquisition in all participants and repeated during stress perfusion acquisition in one participant as proof-of-principle. Post-processing denoise filtering removed gradient artefact and a subject-specific matrix was applied to reconstruct 12-lead ECGs from the raw signal. The morphology of each beat of the in-bore 12-lead ECG was compared to the reference ECG (outside the scanner) using Pearson's correlation coefficient on the PQRST waveform. Results 20 participants were prospectively recruited (60% male, 76±0 years). In-bore 12-lead ECGs were safe and successfully reconstructed in all cases (3 examples at rest: figures 2A-C). The reconstructed 12-lead ECGs (red lines) correlated closely with the standard ECGs (green lines)- mean correlation coefficient r=0.86 (95% confidence interval 0.82;0.90) when comparing PQRST morphology. The mean difference (MD) in QRS duration between ECGs was 4 ms (limit of agreement [LOA] -17 to 25 ms) and MD in cQT interval 1 ms(LOA -22 to 21 ms). ECG recording during adenosine infusion proved feasible in the exemplar (figure 2D- green line, same participant as figure 2C). Recording at peak stress (after 4 minutes of adenosine) showed evolving ST elevation in leads II, III, aVf, V1 and V2 and ST depression in I and aVl. Conclusion High-quality diagnostic 12-lead ECGs can be collected in real-time during a 3T CMR scan with potential to improve triggering and image acquisition. Our ability to detect transient ischaemic changes during stress perfusion may enhance the interpretation of quantitative perfusion maps and the technology may have several neurocardiology applications. Conflict of Interest No