Atrial fibrillation: how to move forward in this complex arrhythmia

Nele Vandersickel

Atrial fibrillation

• Not much progress since famous paper of Haïssaguerre
  PVI isolation as therapy for paroxysmal AF
   

Atrial fibrillation

Different hypothesis in the past 25 years: Each hypothesis has support from models!                                            

Atrial fibrillation

How to move forward?

1. Complexity of models

Models become increasingly complex

1. Advanced Electrophysiological models

Human atrial models

Human ventricular models

Notice the big differences!

Models become increasingly complex

2. Detailed Anatomical representations

• MRI CT scans: highly detailed, patient-specific anatomical models of the heart
• Fiber orientation
• Fibrotic tissue
• Fat
• ...

3. Coupling with mechanics

Heijman et al., Card Res 2021

What if you make the model again as simple as possible? Test for AT!

Direct clinical relevance: topological ablations are being performed in Bruges

131 MRAT cases

Vandersickel et al, 2024, EHJ

Duytschaever et al, 2024 under review

Index theorem on AF

First results show that also for AF: loops come in pairs of 2!

2. Open Source Code

Open source code

• For simulations there are some open source solutions
  
  
  
  
  
  
  
   
• Analysis of electroanatomical data:
  
  
  
   
• Anatomical models, mesh generation
  • Atrial models at scale from imaging or EAM data: Roney, Niederer
  • statistical shape models: Loewe group
  • ...

see also: https://github.com/lileitech/Awesome-Cardiac-Digital-Twins

Open source code

Yet many groups still develop in house code

Field of neuroscience is far ahead of our field:

Platform for neuroscience researchers with the amount of data/software/solutions and tutorials for the community.

Platform for publishing code and data together with providing user with cloud computing resources for pipeline building using this code and data

Datsets, atlases, open software for neuroscience problem.

Open Source seems to be the norm

3. Sharing of datasets

In cardiac research

• Simulations

• Clinical data: very little freely available

Many datasets on the website

...

Neuroscience

Again, neuroscience is far ahead of our field

Platform to share specific data format with the neuroscience community. Implements FAIR principles in data management.

Platform for neuroscience researchers with the amount of data/software/solutions and tutorials for the community.

Platform for publishing code and data together with providing user with cloud computing resources for pipeline building using this code and data

Datsets, atlases, open software for neuroscience problem.

4. Heterogeneous disease

Atrial fibrillation

https://mox.polimi.it/precision-medicine-to-control-atrial-fibrillation-progression/

Also the population might be very heterogeneous!

Take home message

1. Model complexity
   • How complex should our models be to find new mechanisms
   • How correct are our basic models?
     
      
2. Open Source
   • Open source needs to be more mainstream and organized
     
      
3. Data sharing
   • Long way to go to improve our data sharing practices, so we can check each others findings
     
      
4. Heterogeneous patient populations
   • AF probably has many different mechanisms

Directed Graph Mapping: tool to analyze EAM data

New version open source/source available with restrictions for commercial use

You can create your own pipeline!

www.dgmapping.com

Core Team

Nele Vandersickel

Sander Hendrickx

Robin Van den Abeele

Arthur Bezerra

Bjorn Verstraeten

Arstanbek Okenov

Timur Nezlobinskii

Eike Wülfers

Sebastiaan Lootens