Research Groups
Cognitive Computational Neuroscience
This group explores the impact of advances in computing power and machine learning theory to model and study the human brain.
URL:
Contact:
Computer Vision
The computer vision group deals with general problems of detecting structures in images. Particular topics currently include color & reflectance, image forensics, multispectral imaging, multi-camera setups and range imaging. Our work is closely related to other main fields in computer vision, like image segmentation and tracking. Particular topics like image forensics connect closely to statistics, color & reflectance serves often as a pre-processing step for higher level computer vision tasks like object recognition.
URL:
Contact:
Data Processing for Utility Infrastructure
Enterprise Computing
The research group Enterprise Computing (german: Unternehmensinformatik) investigates all aspects of commercial computer applications, in particular distributed applications to process considerably amounts of data. Research and teaching fields are the Mainframe and its linked techniques like transaction processing, virtualization or web applications. In general we concentrate on doing research on planning, development, implementation, installation and security of IT-infrastructures in enterprises and education of the users.
URL:
Contact:
Generative AI
The Generative AI in Medicine (GenAI) group is at the forefront of deep learning research, tackling a wide range of topics involving data types such as image, text, and speech. Our work spans large language models, diffusion models, and GANs to ultimately develop foundation multimodal models. With a particular focus on medicine and healthcare, our multidisciplinary team of engineers and healthcare professionals collaborates to advance AI-driven solutions for diagnosis and prognosis.
URL:
Contact:
Image Analysis
The Image Analysis Group is dedicated to extract information from images. Examples are the outlining of specific structures in 2D and 3D images, like extraction of pages in CT scans of books or the detection of lesions in mammographic images.
URL:
Contact:
Image Fusion
In close collaboration with leading clinical and industrial partners, the Image Fusion (IMF) Group develops novel methods for rigid and non-rigid data registration, as well as innovative applications and efficient clinical workflows. Current foci of interest include multi-modal image fusion, image-guided therapy, 2-D/3-D registration and image overlay. The interdisciplinary research provides the basis to develop applications close to medical practice.
URL:
Contact:
Inverse Problems and Applications (IPA)
In general, inverse problems are concerned with (1) reconstructing signals from observations and/or (2) controlling systems to a desired effect. Inverse problems occur in a variety of domains and applications, ranging from tomographic reconstruction to particle physics to machine learning. The IPA group is dedicated to identifying and solving such inverse problems with a strong focus on radiological applications:
• Tomographic reconstruction for different modalities under non-optimal conditions
• Rigid and non-rigid motion estimation and correction
• Image quality
• Blood flow analysis
• Ionizing radiation dose estimation and optimization
URL:
Contact:
Learning Algorithms for Medical Big Data Analysis (LAMBDA)
This group is concerned with applying the most advanced learning approaches on multi-modal, medical imaging for the improvement of clinical decision making. Current topics of interest include identification of a malignant tumor sub-types in breast cancer, establishing correlations between image-based features, gene expression and disease progression in patients, and developing innovative therapeutic approaches such as immune cell guidance and response activation.
URL:
Contact:
Magnetic Resonance Imaging (MRI)
MRI, or Magnetic Resonance Imaging, is a non-invasive medical imaging technique that provides detailed pictures of the body's internal structures using magnetic fields and radio waves.At our research group, we are committed to advancing MRI technology and its applications. Our research covers a wide range of aspects within MRI, including but not limited to reconstruction, segmentation, and artifact reduction. Through cutting-edge algorithms and methodologies, we aim to improve image quality, enhance anatomical delineation, and reduce unwanted artifacts.
URL:
Contact:
Ophthalmic Imaging
In the recent decade, ophthalmic imaging has proven to be a steadily growing field of research. A milestone in ophthalmology was the transfer of Optical Coherence Tomography (OCT) from research to clinical use. Compared to conventional fundus photography, OCT allows the three dimensional, depth resolved visualization of the human retina, while preserving the non-invasiveness. Recently, a radical change occured in the field of OCT research with the clinical introduction of OCT angiography (OCTA), which further adds dye-free imaging of the underlying vasculature.
Research Focus
To facilitate an efficient analysis of this vastly increased amount of information, new processing algorithms are required to support the treating clinician. Our research focus is twofold: On the one hand, we develop advanced motion compensation, shadow artifact compensation and signal reconstruction algorithms to achieve artifact-free OCT(A) signals of best possible quality. On the other hand, we aid accurate image analysis by improving layer and vessel segmentations, categorizing vessels into arteries/veins or pathology. By combining both efforts, we want to improve the understanding of the most prevalent eye diseases, allowing for more accurate treatment and thus improved patient outcome.
Multidisciplinary Collaborations
To enable research with state-of-the-art technology while preserving a close link to the clinical needs, the work of our group is embedded in a multidisciplinary environment including optical engineers at the Massachusetts Institute of Technology, Cambridge, USA and clinicians at the New England Eye Center, Boston, USA and the Department of Ophthalmology at the University Clinic Erlangen.
Advanced shadow artifact removal reveals the unique structures of the superficial vascular plexus (SVP), the intermediate capillary plexus (ICP) and the deep capillary plexus (DCP)
URL:
Contact:
Precision Learning
Precision Learning is a research direction, seeking to integrate known operators into machine learning models to improve generalization und efficiency.
Known operators have been shown to hold the potential of reducing maximal error bounds when incorporated into deep neural networks. This suggests their inclusion could allow models to learn from less data and increase robustness.
URL:
Contact:
Privacy-Preserving Deep Learning
The Privacy-Preserving Deep Learning (PPDL) group is dedicated to advancing techniques that ensure the confidentiality and security of sensitive data in deep learning applications. Our research focuses on developing and implementing methods such as federated learning, differential privacy, anonymization, and multi-party computation. We work with a diverse range of data types, including images, speech, and text, with a particular emphasis on medical data. Our goal is to create robust and secure AI solutions that protect patient privacy while enabling cutting-edge medical research and healthcare innovations. Through our multidisciplinary approach, we aim to set new standards for privacy in AI and contribute to the ethical advancement of technology in medicine.
URL:
Contact:
Speech Processing and Language Understanding
Our research focuses on modeling speech and language patterns using machine and deep learning methods. We develop spoken dialogue systems, enhance speech, and process out-of-vocabulary words. We analyze prosodic features such as accents and phrase boundaries, and automatically recognize emotion-related states using multi-modal data, including facial expressions, gestures, and physiological parameters. We also recognize user focus in human-machine interactions and analyze pathological speech from children with cleft lip and palate or patients with speech and language disorders. In natural language processing, we develop and apply methods like Large Language Models (LLMs), topic modeling, and part-of-speech tagging, with applications in both medical and industrial domains. We also leverage LLMs and deep learning for advanced speech and language understanding, addressing ethical AI, text summarization, and question/answering systems. Additionally, our work extends to analyzing animal speech (e.g., such as the one from orcas) aiming to interpret communication patterns in zoos and the wild.
URL:
Contact:
Time Series Intelligence
The goal of the Time Series Intelligence (TSI) chair is to use machine learning and artificial intelligence techniques to analyze the complex temporal patterns that can be harnessed to uncover valuable insights, make accurate predictions, and facilitate informed decision-making across a wide range of domains such as radar-based systems, biomedical signals, sensor data, and energy system-based signals. The members of TSI collaborate to address fundamental challenges in time series analysis/signal processing and develop innovative solutions for forecasting and prediction in practical applications, pattern recognition to detect anomalous behavior, classify patterns, and understand their underlying causes, and applying our research findings to real-world problems by working with experts in fields such as healthcare and energy systems to address complex challenges and contribute to advancements in those domains.
URL:
Contact:
Colloquia timetable
For a summary of scheduled colloquia, please refer to the colloquia timetable.