Navigation

Elisabeth Hoppe, M. Sc.

  • Job title: Researcher
  • Organization: Department of Computer Science
  • Working group: Chair of Computer Science 5 (Pattern Recognition)
  • Phone number: +49 9131 85 27874
  • Fax number: +49 9131 85 27270
  • Email: elisabeth.hoppe@fau.de
  • Website:
  • Address:
    Martenstraße 3.
    91058 Erlangen
    Room 09.132

Education

  • Since 10/2017:
    Researcher/PhD Candidate at Pattern Recognition Lab
  • 04/2015 – 08/2017:
    Student at Friedrich-Alexander-Universität Erlangen-Nürnberg, Computer Science (Master of Science)
  • 10/2011 – 03/2015:
    Student at Ostbayerische Technische Hochschule Regensburg, Medical Computer Science (Bachelor of Science)

Professional Employment

  • Since 10/2017:
    Researcher, Siemens Healthineers (Magnetic Resonance), Erlangen
  • 12/2016 – 06/2017:
    Master student, Siemens Healthineers (Magnetic Resonance), Erlangen
  • 09/2016 – 11/2017:
    Internship, Audi Electronics Venture, Gaimersheim
  • 06/2015 – 06/2016:
    Working student, Fraunhofer Institute for Integrated  Circuits IIS, Erlangen
  • 01/2014 – 03/2014:
    Working student, Institut für Vorsorge und Finanzplanung, Altenstadt a.d.Waldnaab
  • 08/2013 – 12/2013:
    Internship, Fraunhofer Institute for Integrated  Circuits IIS, Erlangen
  • 10/2012 – 01/2013:
    Teaching assistant, Ostbayerische Technische Hochschule, Regensburg

 

Gender Equality

My ongoing project is about inspiring young girls for technical and computer science and gender equality. Get yourself informed, why this is so important.

There are plenty events supported by us: Schnupperuni, Schülerinfotag, Mädchen-und-Technik Tag, Girls’ Day, Forscherinnencamp. Please contact me, if you need further information.

3-D Multi-contrast Cardiac CINE Magnetic Resonance Imaging

Research project in cooperation with Siemens Healthineers, Erlangen

Magnetic resonance imaging (MRI) is a non-invasive imaging technique which is well suited for the diagnosis and monitoring of cardiovascular diseases because of its ability of visualizing the anatomy and the functional information of the heart. Additionally, with this technique a diversity of image contrasts is provided. However, cardiovascular MRI is challenging due to e.g. myocardial contraction and respiratory motion and thus not well-established for the clinical practice yet.

With iterative reconstruction methods, the acquisition time can be clearly reduced and the artifacts minimized at the same time. With the help of these methods a representation of the heart with a well spatial and temporal resolution (4-D representation) can be created.

Additionally, quantitative representation of physical relaxation times can be generated with so-called mapping techniques based on these different image contrasts. The aim of this PhD project is the extension of the temporal 3-D representation imaging technique for the heart with such a multi-contrast dimension. This extra dimension can lead to an enhanced separation between pathological and healthy myocardial tissues.

Deep Learning-based Cardiac Navigation for Continuous Cardiac Magnetic Resonance Imaging

Research project in cooperation with Siemens Healthineers, Erlangen

In order to resolve the imaged heart into multiple dynamic dimensions, e.g., respiration and cardiac phases, the data is acquired continuously and afterwards synchronized with the underlying motion within the reconstruction framework. For the cardiac motion, often an external device, e.g., an electrocardiogram (ECG) has to be placed on the subject and monitors the cardiac cycles. However, this additional device is error-prone due to the application within an MR scanner. Further, the workflow is more complicated and the overall scan time is prolonged. We aim at developing a deep learning-based cardiac navigation, which directly derive a specific timepoint during a cardiac cycle. Acquired imaging data is fed into a deep neural network classifier, which outputs the probabilities for an R-wave at every timepoint within the measurement. Using the detected R-waves, data can be binned into different cardiac phases for the dynamic reconstruction. This way, we eliminate the need for an external ECG-device and can simplify the overall workflow.

Deep Learning-based Magnetic Resonance Fingerprinting Reconstruction

Research project in cooperation with Siemens Healthineers, Erlangen

Magnetic Resonance Fingerprinting is a recently proposed quantitative imaging technique. By altering sequence parameters at every time point, different imaging constrasts can be acquired resulting in so-called fingerprints for every image position. These fingerprints are characteristic for the underlying tissue states and can be used to determine quantitative parameters, e.g. T1 and T2 relaxation times. Convential reconstruction methods use pattern matching methods and compare the measured fingerprints with a simulated base of possible fingerprints. However, these methods are inefficient in terms of time and storage and furthermore, can yield errors if the simulated base is too small. To overcome these limitations, we are working on deep learning based MRF reconstruction. Acquired fingerprints are used as inputs for a deep neural network, which directly predicts the quantitative parameters without the need for a comparison with a simulated data base.

 

2020

Journal Articles

Book Contributions

Conference Contributions

2019

Journal Articles

Book Contributions

Conference Contributions

2018

Conference Contributions

2017

Conference Contributions

No patents found.

2020

2019

2018

2017

2019

  • : KI Newcomerin im Bereich Lebenswissenschaften (Gesellschaft für Informatik e.V.) – 2019

2018

  • : conhIT Nachwuchspreis für praxisorientierte Abschlussarbeiten (2. Platz Master Thesis) – 2018

2015

  • : conhIT Nachwuchspreis für praxisorientierte Abschlussarbeiten (1. Platz Bachelor Thesis) – 2015