Seminar: Visualization and Medical Image Analysis


  • Lecturer(s):
  • Start: 03.11.2020 (online)
  • Dates: by arrangement
  • Course number: MA-INF 2219 - Seminar Visualization and Medical Image Analysis
  • Curriculum: Master , B-IT Master Media Informatics, B-IT Master Life Science Informatics
  • Requirements: Knowledge from one of our lectures is recommended.
  • Exams: TBA


In this seminar, you will read and present state-of-the-art papers in visualization or medical image analysis. If you are interested in one of the topics below, please contact the corresponding direct advisor. If you would like to propose your own topic within the scope of visualization and medical image analysis, please contact Prof. Schultz. MSc students in Life Science Informatics can take this seminar, but their topic should be clearly related to Life Science Informatics - please contact us about this if in doubt.

Please use our LaTeX template or a similar layout for your seminar report.

Available Topics

Generative Adversarial Networks for Medical Image Domain Adaptation
Imaging the same anatomical structures with different medical scanners often leads to a different appearance, or different image characteristics. This domain shift can lead to a drastic reduction of accuracy when applying neural networks that have been trained on images from one scanner to those from other devices, or with slightly different settings. It is the goal of this seminar to get an overview of approaches that perform image to image translation with Generative Adversarial Networks (GANs) to address this challenge.
Main reference: Xie et al., MI2GAN: Generative Adversarial Network for Medical Image Domain Adaptation Using Mutual Information Constraint. In: MICCAI, LNCS 12262, pp. 516-525, 2020. (Paper on arXiv)
Direct advisor: Rasha Sheikh

Projective Latent Interventions for Understanding and Fine-Tuning Classifiers
Dimensionality reduction is a common strategy for visualizing latent representations that have been learned in hidden layers of deep neural networks. This work aims to convey a deeper understanding of those projections, and to build better classifiers, by allowing a human user to modify the network by manipulating points in those embeddings.
Main reference: Hinterreiter et al., Projective Latent Interventions for Understanding and Fine-Tuning Classifiers. In: Interpretable and Annotation-Efficient Learning for Medical Image Computing, LNCS 12446, pp. 13-22, 2020. (Paper on arXiv)
Direct advisor: Rasha Sheikh

Relationship between brain structure and function
The relationship between brain structure and function has been one of the centers of research in neuroimaging for decades. In recent years, diffusion tensor imaging (DTI) and functional magnetic resonance imaging (fMRI) techniques have been widely available and popular in cognitive and clinical neurosciences for examining the brain's white matter (WM) micro-structures and gray matter (GM) functions, respectively. Given the intrinsic integration of WM/GM and the complementary information embedded in DTI/fMRI data, it is natural and well-justified to combine these two neuroimaging modalities together to investigate brain structure and function and their relationships simultaneously.
Main reference: Zhu et al., Fusing DTI and fMRI data: A survey of methods and applications. NeuroImage 102(1):184-191, 2014
Direct advisor: Mohammad Khatami

Signal dropout detection and correction in diffusion MRI
Diffusion MRI enables the non-invasive investigation of tissue micro-structure. However, it suffers from acquisition artifacts such as the signal dropout caused by bulk motion during the diffusion encoding part of the imaging. In addition to understanding the paper below, a strong seminar on this topic would involve a literature search on this problem and alternative approaches.
Main reference: Jesper et al., Incorporating outlier detection and replacement into a non-parametric framework for movement and distortion correction of diffusion MR images. NeuroImage 141:556-572, 2016
Direct advisor: Ikram Jumakulyyev

Guidelines for a strong seminar

To successfully finish a seminar in our group, you have to produce two main results:

  • A written report that should be 5-10 pages in length. Strong seminar reports tend to make use of most of the available space, but you should not exceed the limit. For certain topics, such as when theory is involved that is difficult to understand, but can be expressed succinctly, even a report at the shorter end of the range can qualify as strong. We recommend writing the report in LaTeX, since that is a useful skill for the MSc thesis; however, this will not affect your grade.

  • An oral presentation that should take 30 minutes, followed by answering questions about the topic of your seminar. The presentation should contain an introduction that clearly explains what your topic is about and why it is important. Fellow students who have a similar background as you, but did not read the same papers, should be able to follow your presentation. A strong presentation makes use of the available time, but does not exceed it – in extreme cases, we have to cut you off. Your presentation should be supported by suitable media, such as projected slides. It is up to you how you create them (LaTeX, PowerPoint, other means). If it fits your topic, feel free to use other media also, such as the whiteboard.

The written report has to be submitted before the presentation; we will tell you the exact dates well in advance. It is strongly recommended that you discuss your report with your direct advisor – who is usually one of our PhD students – and perform a practice presentation. Please contact them about this well in advance. If you would like to get direct feedback from Prof. Schultz, please submit a draft at least two weeks in advance.

As a basis for your seminar, we will give you a topic and one or multiple scientific publications that are related to it. A basic learning goal is that you should be able to explain, in your own words, the main results reported in those publications. If you would like to cite one of your sources verbatim, you have to clearly mark which part of the text has been copied, and from which source. As a general guideline, such direct citations should only be used if not just a thought or fact in itself, but also the exact way in which the authors expressed it is important. In our discipline, this is relatively rare. If you copy text from papers or other sources, such as Wikipedia or other online material, without clearly marking it, this will be considered as plagiarism and can lead to failing the module!

You may find helpful images, plots, or diagrams in the scientific literature or on the web. It is acceptable to use those, as long as you clearly indicate from which source you took them. It will earn you bonus points if you create your own figures that support your report or presentation, but originality and effectiveness matter more than pure effort in creating them. If you re-create a figure that is very similar or clearly inspired by one that you saw elsewhere, please indicate where the original can be found.

In addition to the papers we give you, a strong seminar requires you to do your own literature search for related publications, so that you will be able to better explain how the paper you are discussing differs from older publications, or how it has influenced more recent ones. Most scientific papers are written for an expert audience, so it is not unusual that you have to look for additional information in the paper’s references or other sources to fully understand it. All sources that were important for creating your final report should be listed in a “References” section at the end, and the text should cite each entry in a way that makes it clear why it is important for your topic.

A thorough literature search will help you to get different perspectives on your topic. However, merely paraphrasing the given text(s) sentence for sentence does not result in a strong seminar. To achieve an excellent grade, it should become clear that you not only understood the papers, but that you are able to abstract, compare, and focus on what is important in your context. Ideally, reflecting on your topic will lead to an independent opinion on aspects such as strengths and weaknesses of currently available methods and reported experiments, different use cases in which different approaches might be preferred, or future research directions.