SwissCardIA


Cardiovascular diseases remain the leading cause of death worldwide. New, innovative approaches are needed to drive progress in this field. This has led to the genesis of the SwissCardIA collaboration, focusing on the use of artificial intelligence (AI) to improve the diagnosis, evaluation, and management of cardiovascular disease.

We believe that progress in this area requires a broad expertise to tackle the complexity of the tasks. SwissCardIA brings together a multidisciplinary research team with expertise in three key domains:

  1. Machine learning (ML) and data science (DS) (EPFL)
  2. Numerical modelling (EPFL)
  3. Clinical cardiology (CHUV, Unisante)

Moreover, SwissCardIA benefits from the support of the Center for Intelligent Systems (CIS) at EPFL.

Latest News

  • First published study demonstrating the potential of AI to predict heart attacks using coronary angiography images, appearing in the journal Open Heart [1].
  • Two more papers in MedNeurIPS:
    • “Anatomy-informed multimodal learning for myocardial infarction prediction” [2]
    • “Attention-based learning of views fusion applied to myocardial infarction diagnosis from x-ray CT” [3]]

Publications and Reports

  1. Mahendiran T, Thanou D, Senouf O et al. Deep learning-based prediction of future myocardial infarction using invasive coronary angiography: a feasibility study. Open Heart 2023;10:e002237. doi: 10.1136/openhrt-2022-002237
  2. I.-D. Sievering, O. Senouf, T. Mahendiran, D. Nanchen, S. Fournier, O. Muller, P. Frossard, E. Abbé, and D. Thanou. Anatomy-informed multimodal learning for myocardial infarction prediction, NeurIPS workshop for Medical Images, Dec., 2022.
  3. J. Gwizdala, O. Senouf, D. Auberson, D. Meier, D. Rotzinger, S. Fournier, S. Qanadli, O. Muller, P. Frossard, E. Abbé, and D. Thanou. Attention-based learning of views fusion applied to myocardial infarction diagnosis from x-ray CT, NeurIPS workshop for Medical Images, Dec., 2022.
  4. D. Thanou, O. Senouf, O. Raita, E. Abbé, P. Frossard, F. Aminfar, D. Auberson, N. Dayer, D. Meier, M. Pagnoni, O. Muller, S. Fournier, and T. Mahendiran. Predicting future myocardial infarction from angiographies with deep learning, NeurIPS workshop for Medical Images, Sept., 2021.


Members


Professor Emmanuel Abbé (ML and DS)
Dr Denise Auberson (Cardiology)
Dr Edward Andò (Center for Imaging)
Professor Annalise Buffa (NM)
Adjunct Professor Simone Deparis (NM)
Dr Fabio Marcinno (NM)
Dr Stephane Fournier (Cardiology)
Professor Pascal Frossard (ML and signal processing)
Dr Thabo Mahendiran (Cardiology)
Professor Olivier Muller (Cardiology)
Dr David Nanchen (Cardiology)
Dr Mattia Pagnoni (Cardiology)
Dr Ortal Senouf (ML and DS)
Dr David Rotzinger (Radiology)
Dr Ricardo Tenderini (NM)
Dr Dorina Thanou (ML and signal processing)

Current projects

1. Prediction of future myocardial infarction (MI)

A major goal of the collaboration is the application of AI to improve the prediction of MI and other adverse patient events. Given the complexity of the prediction task, our approach involves the integration of multiple data sources:

  • Cardiac imaging with a particular focus on Invasive coronary angiography and CT coronary angiography
  • 3D coronary reconstructionto better incorporate the anatomical drivers of CAD
  • Numerical modelsthat harness 3D models to compute the haemodynamic drivers of coronary artery disease, such as fractional flow reserve (FFR) and wall shear stress (WSS)

2. AI4HealthyCities: improving the cardiovascular health of Lausanne, Switzerland

This collaboration between EPFL, Unisanté-CHUV, Direction générale de la santé (Vaud), and the Novartis Foundation will aim to improve the stratification of cardiovascular disease (CVD) risk in Lausanne. In particular, our approach focuses on health inequalities and integrates the social determinants of health - the economic, social, environmental, and psychosocial factors that influence health. These are important drivers of CVD that are not accounted for in current mainstream risk calculators.