Influence of the MHD effect on 12-lead and 3-lead ECGs recorded in 1T to 7T MRI scanners

Abstract

This ECG dataset was acquired in different magnetic resonance imaging (MRI) scanners to study the magnetohydrodynamic (MHD) effect. The MHD effect, which is caused an interaction of the MRI’s strong static magnetic field and the patient’s blood flow, superimposes the ECG signal during MRI exams. As a consequence, a detailed morphological analysis of the ECG (e.g. of the P wave, ST segment or the T wave) is prevented. The MHD effect might be a useful signal for extracting further physiological information, e.g. about blood flow or stroke volume. The 12-lead and 3-lead ECG signals were acquired from different subjects in various MRI scanners with magnetic field strengths ranging from 1T up to 7T. All signals were manually annotated.

Background

During an MRI exam, a subject or patient is exposed to a strong static magnetic field which typically ranges from 1T to 3T in clinical scanners and up to 10.5T in research scanners. The physical interaction between the static magnetic field and the pulsatile blood flow, which is caused by the rhythmic action of the heart, results in the magnetohydrodynamic (MHD) effect.

Ions (electrolytes) contained in the blood are moving inside the vessels where they experience a force due to the presence of the MR scanner’s static magnetic field. This force, which is known as Lorentz force, causes the ions to move perpendicular to the direction of the blood flow and perpendicular to the MR scanner’s static magnetic field. The ions accumulate near the vessel’s wall leading to a potential difference across the vessel. This potential difference or voltage is referred to as the Hall voltage. The Hall voltages across the blood vessels lead to blood flow dependent body surface potentials which are superimposing the ECG signals.

Due to the superposition of the ECG and MHD signals, a detailed and reliable morphological analysis of the ECG during MRI exams, e.g. of the P wave, ST segment or the T wave, is not possible [1,2,6,9,10]. Another challenge is the detection of the QRS complex. Depending on the characteristics of the MHD signal, QRS detection might be hampered [3]. Several dedicated QRS detection algorithms were developed to cope with this issue [4-8].

One potential goal for using this dataset would be to increase the diagnostic value of an ECG acquired during MRI exams, which would be especially important for critical care patients or during MRI guided interventions. This goal could be achieved by improving existing signal processing techniques aiming to separate ECG and MHD signal components.

Due to the high correlation between the ECG and MHD signals and overlapping frequency distributions, a separation of both signals still remains a challenging task to solve. Another difficulty is the lack of a reference ECG while the subject is inside the MRI scanner, which complicates the validation of the developed algorithms. To cope with this problem, reference ECGs were acquired outside the scanner assuming stationarity of the ECG signals for healthy subjects without cardiovascular diseases or arrhythmias.

Methods

ECGs were acquired in different MRI scanners at Otto-von-Guericke University of Magdeburg, Germany, with static magnetic field strengths of 1T, 3T and 7T without imaging, i.e. in the absence of the switched gradient and the high frequency magnetic fields. Hence, the ECG signals are only distorted by the MHD effect but not by the switched gradient magnetic fields or high frequency magnetic fields which are present during active MR imaging.

ECG Hardware and MRI scanners

ECG signals were acquired using two different ECG recorders:

1. a 12-lead Holter ECG (CardioMem 3000,Getemed AG, Germany) with a sampling rate of 1024 Hz, an input voltage range of +/-6mV, a resolution of 12 bit and an analogue bandwidth of 0.05 Hz to 100 Hz.
2. the wireless 3-lead ECG device of an MRI-conditional patient monitoring system (Tesla M3, MIPM GmbH, Germany) with a sampling rate of 1024Hz, an input voltage range of +/-2.4mV and a resolution of 24 bit.

Subjects and data acquisition

The database comprises 43 records from 23 different subjects with an overall length of 203 min. The subjects had an average age of 27.1 +/-3.2 years, an average weight of 73.8 +/- 13.1 kg and a height of 181.7+/-10.5 cm. The subjects, who voluntarily participated in this study, were students and employees from the Otto-von-Guericke University of Magdeburg, Germany. The data was collected in the years 2011-2017 and the study was approved by the local ethics committee.

Most of the datasets were acquired during spontaneous, free breathing. For certain datasets, breath-hold commands were applied asking the subjects to hold their breath for several seconds in order to avoid or reduce respiratory modulations of the ECG and MHD signals. This is noted in the header files. Physicians or ECG experts manually annotated the QRS complexes. No distinction was made between QRS complexes occurring in normal or disturbed heart rhythm, e.g. caused by ectopic beats.

Data Description

All files are provided in standard WFDB format, and the sampling rate is 1024 Hz.

Data format, header and annotation files

The filenames specify the subject number, the magnetic field strength given in Tesla (T), and the measurement position inside the MRI scanner: head first (Hf), feet first (Ff), Pro (prone), Sup (supine) or Out (outside the scanner).

The header file *.hea of each record contains the following information: technical parameters of the MR scanner (field strength, manufacturer, field orientation), ECG hardware parameters and information about the subject (sex, age, weight, height, position in the scanner, type of respiration, ECG details where applicable). For most of the subjects, a reference ECG signal was recorded outside the MRI scanner. QRS complexes were manually annotated for each record and are provided in the *.qrs file.

Magnetic field orientation

There is one point to note with respect to the orientation of the static magnetic field B0 of the MRI scanner. For most scanners available nowadays, the B0 field has a horizontal orientation, i.e. the field points from head to feet (or vice versa) while the subject or patient is inside the scanner. The 1T scanner used for the acquisition of this dataset was a (now discontinued) open field scanner where the B0 field has a vertical orientation pointing from dorsal to ventral (i.e. from back to chest). This has an important influence of the morphology and magnitude of the MHD effect [1,6]. Hence, measurements in the 1T scanner were performed in the prone and supine position (always head first) to invert the MHD effect whereas for all other scanners (1.5T, 3T and 7T), measurements were performed in head first and feet first positions (always supine).

Example file names

• ECGMRI3T04Ff: 3T, Subject #04, Feet first position.
• ECGMRI3T04Out: 3T, Subject #04, Outside scanner (reference ECG).

Usage Notes

The ECG MRI database was created for the purpose of developing and evaluating algorithms for filtering or separating ECG and MHD signals enabling an improved and more reliable patient monitoring during MRI exams. It shall also stimulate the further analysis of the MHD effect for other diagnostic purposes.

The dataset has been previously used by the author in different studies with the purpose the analyze and filter the MHD effect [1,2,6,8] as well as for developing dedicated QRS detection algorithms [3-8].

One limitation of this dataset is that it was acquired from young, heathy subjects without known (cardiac) diseases. Hence, this dataset is not suitable for studying the MHD effect during cardiac arrhythmias or other cardiovascular diseases.

Acknowledgements

The author would like to thank Georg Rose for supporting this research project and providing access to the MRI systems. Special thanks go to Julien Oster and Gari Clifford for the long-standing and fruitful collaboration and for their support in analysing and processing the ECG signals. This project would not have been possible to this extent without the support of the colleagues from the Department of Medical Engineering at OVGU Magdeburg, the clinical partners and collaborators from the University Hospital Magdeburg as well as MIPM. The following research grants allowed the acquisition of the ECG datasets in the years between 2011 and 2017: German BMBF grants 03IP710, 03IPT7100X, 03FO16101A and the German BMWi grant KF3172301JL3.

Conflicts of Interest

The author has no conflict of interest.

References

1. Krug, J. W., & Rose, G. (2011, September). Magnetohydrodynamic distortions of the ECG in different MR scanner configurations. In 2011 Computing in Cardiology (pp. 769-772). IEEE.
2. Krug, J. W., Rose, G. H., Stucht, D., Clifford, G. D., & Oster, J. (2012, September). Filtering the magnetohydrodynamic effect from 12-lead ECG signals using independent component analysis. In 2012 Computing in Cardiology (pp. 589-592). IEEE.
3. Krug, J. W., Rose, G., Stucht, D., Clifford, G., & Oster, J. (2013). Limitations of VCG based gating methods in ultra high field cardiac MRI. Journal of Cardiovascular Magnetic Resonance, 15(1), 1-2.
4. Krug, J. W., Rose, G., Clifford, G. D., & Oster, J. (2013). ECG-based gating in ultra high field cardiovascular magnetic resonance using an independent component analysis approach. Journal of Cardiovascular Magnetic Resonance, 15(1), 1-13.
5. Schmidt, M., Krug, J. W., Gierstorfer, A., & Rose, G. (2014, September). A real-time QRS detector based on higher-order statistics for ECG gated cardiac MRI. In Computing in Cardiology 2014 (pp. 733-736). IEEE.
6. Krug, J. W. (2015). Improved cardiac gating and patient monitoring in high field magnetic resonance imaging by means of electrocardiogram signal processing, PhD Thesis, Otto-von-Guericke University Magdeburg.
7. Schmidt, M., Krug, J. W., & Rose, G. (2016). Real-time QRS detection using integrated variance for ECG gated cardiac MRI. Current Directions in Biomedical Engineering, 2(1), 255-258.
8. Haritopoulos, M., Krug, J., Illanes, A., Friebe, M., & Nandi, A. K. (2017). Cyclostationary analysis of ECG signals acquired inside an ultra-high field MRI scanner. In 2017 25th European Signal Processing Conference (EUSIPCO) (pp. 1300-1304). IEEE.
9. Passand, J. K., & Rose, G. (2019). Progress of MRI-guided EP Interventions is Hampered by a Lack of ECG-based Patient Monitoring-An Engineering Perspective. In BIOSIGNALS (pp. 201-208).
10. Kinouchi, Y., Yamaguchi, H., & Tenforde, T. S. (1996). Theoretical analysis of magnetic field interactions with aortic blood flow. Bioelectromagnetics: Journal of the Bioelectromagnetics Society, The Society for Physical Regulation in Biology and Medicine, The European Bioelectromagnetics Association, 17(1), 21-32.