curAIheart: Artificial intelligence based evaluation of echocardiographic image data considering high dimensional clinical data.

1 PhD position in thecurATime clusters4future initiative offered in IPP winter call 2022

Scientific Background

Echocardiography is a time-consuming, clinically established procedure for determining structural and functional parameters of the heart, which play a crucial role in the classification and risk assessment of patients with cardiovascular diseases. The measurements used in clinical routine today are based on visual changes that can be detected by the human eye in recorded video loops. Newer parameters, such as "global longitudinal strain" (GLS), capture information that is not accessible to the human eye via utilizing image-tracking techniques that allow improved risk stratification of different cardiac patients. However, these new metrics require preselected region of interest. Advances in computational power have enabled the use of machine learning methods, such as Convolutional Neural Networks (CNNs), to analyze medical data.

As part of the large portfolio of clinical studies (including Gutenberg Health Study, MyoVasc) of the Clinical Epidemiology and Systems Medicine (PI: Prof. Dr. Philipp Wild), >20.000 participants underwent extensive deep clinical phenotyping in a highly standardized manner at the University Medical Center Mainz, resulting in a unique and comprehensive dataset. This multi-petabyte dataset includes multi-omics data with state-of-the-art 2D and 3D cardiovascular ultrasound phenotyping. In the curAIheart project, researchers use this unique basis to develop a novel AI-based pipeline for automated echocardiographic image analysis and subsequent biomedical exploitation for translational research in the field of atherothrombosis.

PhD Project: "curAIheart"

This project is part of “CurATime – Cluster for Atherothrombosis and Individualized Medicine” (, a research cluster recently funded by the German Federal Ministry of Education and Research (BMBF) for €15 million for the first 3-year funding period. The goal of the PhD project is to develop innovative technologies and pipelines for the analysis of biodata on cardiac function and structure and to integrate this novel pipeline in a systems medicine oriented approach for biomedical research on the development and progression of atherothrombosis. As part of this patient-oriented translational research project your tasks would include:

  • In close collaboration with Institute for Informatics (Prof. Stefan Kramer, Johannes Gutenberg University Mainz) you will be developing deep learning models for a novel echocardiographic pipeline

  • Facilitating the development of methods that can explain the predictions of a model at the individual and dataset level.

  • You will explore the complex interplay between cardiac structure and function across the span of early and late prevention of atherothrombosis by integrating multi-omics data for generation of an improved disease understanding.

The candidate will be integrated in a friendly, professional and highly multidisciplinary team, comprising clinicians, epidemiologists, bioinformaticians, biostatisticians, as well as biologists and biochemists. Specific competences and supervisors are present to support the PhD candidate. Within the curAIheart project, the candidate will additionally interact with experts in the fields of artificial intelligence (Institute for Informatics, University of Mainz and DFKI, German Research Center for Artificial Intelligence), experimental research (Center for Thrombosis and Hemostasis Mainz) and biotechnology (TRON/BioNTech).

Publications relevant to this project

  • Trobs, S.O.*, Prochaska J.H.* et al., Association of Global Longitudinal Strain With Clinical Status and Mortality in Patients With Chronic Heart Failure. JAMA Cardiol, 2021.

  • Wild PS et al, Large-scale genome-wide analysis identifies genetic variants associated with cardiac structure and function. J Clin Invest 2017.

  • J. Zhang et al., Fully automated echocardiogram interpretation in clinical practice: Feasibility and diagnostic accuracy, Circulation, 2018.

  • D. Ouyang et al., “Video-based AI for beat-to-beat assessment of cardiac function,”Nature, 2020.

  • Madani, R. Arnaout, M. Mofrad, and R. Arnaout, Fast and accurate view classification of echocardiograms using deep learning, NPJ Digit. Med., 2018.

Contact Details

Dr. Jürgen Prochaska (primary contact)
Group leader, Clinical Epidemiology and Systems Medicine, Center for Thrombosis and Hemostasis, Mainz
Deputy Head of Preventive Cardiology and Medical Prevention, University Medical Center Mainz

Prof. Philipp S. Wild (senior supervisor)
Head, Clinical Epidemiology and Systems Medicine, Center for Thrombosis and Hemostasis, Mainz
Head, Preventive Cardiology and Medical Prevention, University Medical Center Mainz
Systems Medicine, Institute of Molecular Biology