Preliminary Validation Using in vivo Measures of a Macroscopic Electrical Model of the Heart

Maxime Sermesant, Owen Faris, Franck Evans, Elliot R. McVeigh, Yves Coudière, Hervé Delingette, Nicholas AyacheIS4TH 2003 · DOI

TL;DR

This paper presents a novel experimental protocol and computational framework to validate macroscopic electrical models of the heart using in vivo measurements. By coupling detailed canine heart electrical activity recordings with sophisticated 3D modeling, the authors pave the way for more accurate predictive simulations of cardiac pathologies and interventions.

What Problem Does This Paper Address?

Understanding and predicting cardiac electrical activity disorders is crucial for diagnosing and treating various heart conditions. While computational models offer a powerful tool for this, their accuracy hinges on validation against real-world physiological data. However, obtaining direct, in vivo measurements of cardiac electrical activity, especially at a macroscopic level, is challenging and invasive. This paper tackles this gap by developing a protocol to acquire such data and using it to validate and refine existing macroscopic heart models, aiming to enable their use in predictive scenarios like planning radio-frequency ablation.

Methods

The authors employ a multi-faceted approach combining experimental data acquisition with advanced computational modeling.

Experimental Protocol: * In vivo measurements: They used a canine heart model, inducing anesthesia and performing a median sternotomy. A custom-designed multi-electrode epicardial sock with 128 copper electrodes was placed over the ventricular epicardium. RV pacing was used to suppress intrinsic electrical activation, allowing for the recording of epicardial electrical potentials. * Data acquisition: Recordings were made at high sampling rates (1000-1450 Hz) and synchronized with MR scans. * Post-mortem processing: After euthanasia, the heart was excised, filled with vinyl polysiloxane to maintain its shape, and then digitized to precisely locate electrodes and anatomical landmarks.

Computational Modeling: * Geometry: Two different models of the myocardium geometry were built: one from a dissected canine heart and another from Diffusion Tensor Imaging (DTI). These models were meshed using tetrahedral elements. * Electrical Model: A simplified Aliev and Panfilov model was used to describe the electrical activity at the cell level. This model was extended to a macroscopic level, incorporating anisotropy derived from fiber directions. * Simulation: The propagation of electrical signals was simulated on the 3D tetrahedral meshes. The simulations incorporated anisotropy, with two different fiber direction datasets (one smoothed, one from DTI) being compared. * Validation: The simulated activation patterns (isochrones) were qualitatively compared with the measured data and existing literature data (Durrer et al.).

Datasets

  • In vivo canine heart electrical activity data: This is a newly acquired dataset from the experimental protocol described in the paper.
  • Cardiac muscle fiber direction data: Two sources were used:
  • Interpolated and smoothed fiber directions from a dissected canine heart (Auckland University).
  • Diffusion Tensor Imaging (DTI) data from a canine heart (Duke University Medical Center).
  • Literature data: Action potential isochrones from Durrer et al. [4] were used for initial validation.

Key Insights and Ideas

  • Bridging the gap between in vivo data and models: The paper successfully demonstrates a protocol for acquiring detailed in vivo electrical data and using it to validate macroscopic heart models, a critical step for clinical translation.
  • Impact of anisotropy: The study highlights the significant influence of myocardial fiber direction on electrical propagation, with the DTI-derived, noisier fiber data showing a more pronounced effect of anisotropy compared to the smoothed model.
  • Model selection for validation: The qualitative comparison suggests that a smoothed model might offer more consistent results when the underlying fiber data is noisy, but future work aims to use more precise data from the same heart for a definitive comparison.
  • Potential for predictive modeling: The ultimate goal of using validated models for predictive simulations of pathologies and interventions (like ablation planning) is clearly articulated.

Key Results

  • Qualitative validation: The simulated activation patterns from both the smoothed and DTI-based models qualitatively matched the measured and literature data.
  • Anisotropy effect: Simulations showed that radial diffusion is significantly slower in anisotropic models, especially when using DTI data, due to discontinuities in fiber directions.
  • Model comparison: The paper suggests that the smoothed model might provide a more robust initial comparison, while acknowledging the need for more precise data to fully evaluate the DTI model.

Limitations and Future Work

  • Qualitative vs. Quantitative Validation: The current validation is primarily qualitative. The authors state that quantitative adjustment of model parameters using data assimilation techniques (like Kalman filtering) is an ongoing work.
  • Model Complexity: The chosen model is simplified to allow parameter adjustment from macroscopic measures. More complex models might not be tractable with the current data.
  • Data Assimilation Challenges: Existing data assimilation methods are not perfectly suited for adjusting such models with the available data.
  • Future Directions:
    • Perform quantitative adjustment of model parameters.
  • Use DTI data from the same heart used for electrical measurements for a more direct comparison.
  • Develop methods to find electrical activity patterns directly from ECG measures.
  • Integrate the electrical model with an electro-mechanical coupling model for a comprehensive simulation of heart function.
  • Utilize validated models for predictive simulations of pathologies and intervention planning.

Summary updated

BibTeX
@inproceedings{sermesant2003preliminary,
  author = {Maxime Sermesant and Owen Faris and Franck Evans and Elliot R. McVeigh and Yves Coudière and Hervé Delingette and Nicholas Ayache},
  title = {{Preliminary Validation Using in vivo Measures of a Macroscopic Electrical Model of the Heart}},
  booktitle = {IS4TH},
  year = {2003},
  doi = {10.1007/3-540-45015-7\_22},
}
Figure