Understanding the Genesis of Cardiac Signals in Electrical Impedance Tomography

Martin Proença, Fabian Braun, Mathieu Lemay, Bartlomiej Grychtol, M. Bührer, Michaël Rapin, P. Krammer, Stephan H. Böhm, Josep Solà, Jean-Philippe ThiranBIOSIGNALS 2014 · DOI

TL;DR

This paper investigates the origins of cardiac signals detected by Electrical Impedance Tomography (EIT), a non-invasive imaging technique. They developed a dynamic bio-impedance model to simulate EIT signals and found that cardiac motion, not just blood volume changes, significantly influences these signals, potentially dominating them by up to 70%. This work is crucial for improving the accuracy and interpretation of EIT in cardiovascular applications.

What Problem Does This Paper Address?

Electrical Impedance Tomography (EIT) is a promising, low-cost imaging technology for monitoring physiological processes within the thorax, such as ventilation. While EIT has been extensively used for respiratory monitoring, its application to cardiovascular signals has gained traction more recently. These cardiac signals, detected as impedance changes, are thought to arise from variations in blood volume within the heart and major vessels. However, there's growing evidence suggesting that the mechanical action of the heart muscle (myocardium) and the movement of the heart-lung interface also contribute to these impedance changes. Disentangling these different sources of signal is critical for accurately interpreting EIT data for hemodynamic monitoring and diagnosis, but it's a complex challenge due to the diffusive nature of electrical currents and anatomical ambiguities in EIT images.

Methods

The researchers tackled this problem by creating a novel dynamic 2D bio-impedance model. This model was built from segmented human magnetic resonance (MR) imaging data acquired over a full cardiac cycle. The key innovation lies in simulating three distinct scenarios:

  • Scenario A (Blood Volume Only): This scenario isolates the effect of changes in blood volume within the cardiac cavities (atria and ventricles) while keeping the myocardium and other tissues static.
  • Scenario B (Motion Only): This scenario focuses solely on the mechanical dynamics of the myocardium (contraction and relaxation) and its movement, keeping blood volume changes constant.
  • Scenario C (Combined): This scenario simulates the combined effects of both blood volume changes and cardiac motion, aiming to represent a more realistic physiological situation.

To analyze the EIT signals generated by these scenarios, they employed the GREIT (Graz Consensus on EIT) algorithm for image reconstruction. This algorithm is known for its ability to encode performance requirements directly into the inverse matrix, making it suitable for accurate EIT simulations. A control scenario (A+B) was also created by summing the results of scenarios A and B to compare with scenario C.

Datasets

The study utilized segmented human magnetic resonance (MR) imaging data. This data was acquired from a single 83-kg, 183-cm tall 50-year-old male subject with a specific underbust girth. The MR images were captured using a 3T Philips Achieva instrument and were ECG-gated and breath-held, performed in an oblique plane along the heart's long axis. A total of 20 2D image slices were acquired, representing a full cardiac cycle with a temporal resolution of 43 ms. This dataset was segmented manually to create the dynamic bio-impedance model.

Key Insights and Ideas

  • Cardiac Motion is a Major Player: The most significant finding is that cardiac motion, independent of blood volume changes, can substantially influence EIT signals in the heart region. This challenges the long-held assumption that EIT cardiac signals are primarily driven by blood volume shifts.
  • Motion Can Dominate: The simulations suggest that motion-induced impedance changes can account for up to 70% of the total cardiac EIT signal in the myocardial and ventricular regions. This highlights the need to account for mechanical effects in EIT analysis.
  • Spatial Variation in Influence: The impact of blood volume and motion varies spatially. Blood volume changes appear to have a stronger effect in the ventricular region, while motion's influence is more pronounced in the myocardial region.
  • Lung Signals are Complex: For the anterior lung pixel, the influence of both blood volume and motion was found to be roughly equal, suggesting that cardiac activity can indeed perturb signals in adjacent lung regions.
  • Model Validation: The study validates the observed impedance waveforms against known physiological events of the cardiac cycle, lending credibility to the model's behavior.

Key Results

  • Simulations showed that in the myocardial (M) and ventricular (V) pixels, the impedance signal is dominated by motion-induced changes (Scenario B).
  • In the anterior lung pixel (L), the influence of both blood volume (Scenario A) and motion (Scenario B) was found to be approximately equal.
  • Figure 3b quantifies the contribution of each factor: for the myocardial and ventricular pixels, motion-induced changes contributed significantly more to the overall impedance change in scenario C compared to blood volume changes.
  • The study demonstrated that the combined effects of blood volume and motion (Scenario C) closely mirrored the sum of their individual effects (Control Scenario A+B), suggesting a largely additive relationship in this model.

Limitations and Future Work

The authors acknowledge several limitations of their study:

  • 2D Model: The model is two-dimensional, whereas electrical currents propagate in three dimensions. This simplification means that impedance changes occurring above and below the electrode plane are not adequately considered.
  • Simplified Physiology: The model does not account for the dynamics of other organs and tissues, such as respiratory movements, pulmonary perfusion, or blood volume changes in major elastic arteries. While this isolation is useful for understanding individual factors, it neglects their real-world interdependencies.
  • Single Subject: The model was based on data from only one subject.

Future work suggested includes:

  • Exploring Higher Frequencies: Investigating larger frequency content or multi-frequency approaches could help in source separation.
  • Motion-Compensating Models: Developing models that can compensate for motion could enhance EIT signal interpretation.
  • Experimental and Clinical Validation: Conducting experimental and clinical studies is crucial to confirm the findings observed in the simulation model.
  • 3D Modeling: Extending the model to three dimensions would provide a more comprehensive representation of current propagation and physiological dynamics.

Summary updated

BibTeX
@inproceedings{proenca2014understanding,
  author = {Martin Proença and Fabian Braun and Mathieu Lemay and Bartlomiej Grychtol and M. Bührer and Michaël Rapin and P. Krammer and Stephan H. Böhm and Josep Solà and Jean-Philippe Thiran},
  title = {{Understanding the Genesis of Cardiac Signals in Electrical Impedance Tomography}},
  booktitle = {BIOSIGNALS},
  year = {2014},
  doi = {10.5220/0004793400270034},
}
Figure