Graph-based data analysis for the interpretation
of atrial arrhythmias

Nele Vandersickel

Graph-based data analysis and topology

for the interpretation
of atrial arrhythmias

Nele Vandersickel

Core Team

Nele Vandersickel

Sander Hendrickx

Robin Van den Abeele

Arthur Bezerra

Bjorn Verstraeten

Arstanbek Okenov

Timur Nezlobinskii

Eike Wülfers

Sebastiaan Lootens

Main Clinical collaborations

Mattias Duytschaever

Sebastiaan Knecht

Armin Luik

Annika Haas

Atrial tachycardia: case study

Looks like typical case of flutter?

Why do slower ATs arise after ablation?

Entrainment mapping finds the driving source

Slower AT after initial ablation, why do they arise?

Anterior

Posterior

Reentry at the roof

Slower AT at the MV

Anterior

Posterior

Roof ablation

Entrainment mapping finds the driving source

Anterior

Posterior

Reentry at the roof

Roof ablation: slower AT at the MV

Anterior

Posterior

Why do slower ATs arise?

  • 1/3 of the cases in our database

Slower AT after initial ablation, why do they arise?

We tried to create the most simple system

MV

LPV

RPV

Anatomy of the left atrium: 3 natural boundaries

MV

LPV

RPV

SVC

IVC

TV

Anatomy of the left atrium: 3 natural boundaries

Simulation on spheres with 3 boundaries

200 different simulations

Entrainment mapping

All possible virtual ablation lines: 600 simulation

3 Patterns

Simulation on spheres with 3 holes

Complete rotation

Incomplete rotation

Parallel activation

Good entrainment

Bad entrainment

Bad entrainment

Ablation

Incomplete rotation becomes complete rotation, resulting in a slower AT

100% of simulations!

Index theorem (> 20 years old!)

 

Indexes

Complete rotation

Parallel activation

Incomplete rotation

Critical Boundary

Critical Boundary

Non-Critical Boundary

CB: Santucci et al. JACC EP 2023

Ablation: connect the critical boundaries

CB:

CB:

NCB: 0

Incomplete rotation becomes complete rotation, resulting in a slower AT

Loops come in pairs of 2: currently second loop is always missed

Clinical cases

What happens in case we only have 2 boundaries?

Reentry around 1 boundary equivalent with reentry around other boundary

200 simulation with 2 boundaries

Clinical cases

Ablation: connect the critical boundaries

CB:

CB:

NCB: 0

Incomplete rotation becomes complete rotation, resulting in a slower AT

More than 3 boundaries: additional scar tissue

Scar creates additional holes

200 simulations with 4 boundaries

Clinical cases

All 131 clinical cases together

Clinical cases

131 MRAT cases

Clinical cases

20 detailed cases with slowing after ablation

Macro reentry

Localized reentry

Micro reentry

Focal

Rotor

Around anatomic obstacle, like valve or vessels.

Around non conducting area > 2-1.5cm, e.g. scar or functional block

  • Atypical Flutter    (LA involving valves or vessels)

  • Atypical Flutter   (RA involving valves or vessels)​

  • Typical Flutter (clockwise and counter clockwise)

  • WPW

  • Atypical Flutter    (LA with scar or previous ablation lines with gaps)​

  • AVNRT    (considering slow  and fast pathway)

Around non conducting area < 1cm, e.g. scar or functional block

  • Atypical Flutter    (LA with scar​)

  • Atypical Flutter   (RA crista terminalis region)

  • Ectopic AT

Spiral activation pattern with no scar in the core.

Focal activation from  one single spot with centrifugal activation pattern

CL tends to be longer >350ms due to larger path

CL range varies and CL can shift around 20-40 ms during arrhythmia due to slight path variations

CL tends to be shorter due to short path, whole CL is covered by signals found in a small area like e.g. 4cm2 often a combination of double potentials in the „middle“ and very long fractionated signals surrounding​

CL often not presented completely in one chamber

Focal activation patterns could also hint to epicardial entries – look for potential conducting structure like vein of Marshall, Bachmann´s etc. and exits that would fit a macro reentry with epicardial parts

We think non-existent in AT, especially as a stable configuration leading to a stable AT.

Maybe something close to a spiral pattern can exist in AF

  • AF

Current Classification...

Macro reentry

Localized reentry

Micro reentry

Focal

Rotor

Around anatomic obstacle, like valve or vessels.

Around non conducting area > 2-1.5cm, e.g. scar or functional block

  • Atypical Flutter    (LA involving valves or vessels)

  • Atypical Flutter   (RA involving valves or vessels)​

  • Typical Flutter (clockwise and counter clockwise)

  • WPW

  • Atypical Flutter    (LA with scar or previous ablation lines with gaps)​

  • AVNRT    (considering slow  and fast pathway)

Around non conducting area < 1cm, e.g. scar or functional block

  • Atypical Flutter    (LA with scar​)

  • Atypical Flutter   (RA crista terminalis region)

  • Ectopic AT

Spiral activation pattern with no scar in the core.

Focal activation from  one single spot with centrifugal activation pattern

CL tends to be longer >350ms due to larger path

CL range varies and CL can shift around 20-40 ms during arrhythmia due to slight path variations

CL tends to be shorter due to short path, whole CL is covered by signals found in a small area like e.g. 4cm2 often a combination of double potentials in the „middle“ and very long fractionated signals surrounding​

CL often not presented completely in one chamber

Focal activation patterns could also hint to epicardial entries – look for potential conducting structure like vein of Marshall, Bachmann´s etc. and exits that would fit a macro reentry with epicardial parts

We think non-existent in AT, especially as a stable configuration leading to a stable AT.

Maybe something close to a spiral pattern can exist in AF

  • AF

All of this is replaced by our simple classification!

Current Classification...

2nd Submitted paper

Atrial topology for a unified understanding of typical and atypical flutter

Mattias Duytschaever, Robin Van Den Abeele ... Annika Haas, Armin Luik, ... Sander Hendrickx, Nele Vandersickel

European Heart Journal

Atrial tachycardia: case study

Atrial tachycardia: case study

How to find the critical boundaries automatically?

Directed Graph Mapping

Tool that can analyze electro-anatomical mapping data using network theory

www.dgmapping.com

Directed Graph Mapping

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

You can create your own pipeline!

www.dgmapping.com

Directed Graph Mapping: features

www.dgmapping.com

  • Input:
    • OpenCARP
    • RHYTHMIA
    • CARTO
    • Optical mapping data                                            
  • Calculation:
    • Pre-processing
    • Many different phase mapping methods
    • Graph methods and measures
    • Reentry loops (complete and incomplete)
    • Post-processing
  • Visualization:
    • Extended toolkit for visualization
  • Endlessly adaptable

Many examples!

Full Documentation

Tutorials

Future research

  • Bi-atrial
    • More complex topology
    • Voids

 

  • Atrial fibrillation
    • Functional reentry
    • Mathematical boundaries (singularities)

 

  • Ventricles: thickness
    • 2D manifolds

Take home message

DGM: diagnostic tool based on network theory

CB:

CB:

NCB: 0

Incomplete rotation becomes complete rotation, resulting in a slower AT

Topology: Paired loop paradigm

Simplifying can help you find new things even in 'simple arrhythmia'!

www.dgmapping.com