Devesh Singh | German Center for Neurodegenerative Diseases (DZNE), Rostock & University Medical Center (UMR), Rostock

About Me

I’m a Ph.D. researcher at the University Medical Centre Rostock (UMR) and the German Center for Neurodegenerative Diseases (DZNE) in Rostock. I recently began supporting UMR in its implementation of the Open Medical Inference (OMI) platform to enable peer-to-peer exchange of healthcare data and access to remote AI services. Previously at DZNE, I was working within the Clinical Dementia Research Group under the guidance of Prof. Dr. Stefan Teipel and Dr. Martin Dyrba. In 2022, I graduated with an M.Sc. in Data and Knowledge Engineering from Otto von Guericke University (OvGU), Magdeburg, where I specialized in deep learning and computer vision.

My research focuses on building explainable AI models for disease detection using brain MRI data. I’m particularly interested in making deep learning systems more interpretable and clinically useful. One of my projects focused on developing self-supervised learning methods for analyzing brain scans and evaluating how well these models capture imaging features sensitive to neurodegenerative diseases. I’ve also worked on a semantic learning task where an ontology-based framework helped organize and visualize brain atrophy patterns, learning hierarchical rule-based features to make complex pathologic information more accessible. Currently, I’m co-supervising a Master’s thesis investigating a hybrid explainability approach combining rule-based reasoning with Large Language Models (LLMs) to generate clear, clinically relevant explanations from MRI-derived metrics. Outside of academia, I’ve had the opportunity to work with STIHL, SMA Solar Technology, and SICK in student roles, where I applied AI to real-world problems. All these experiences continue to shape my passion for building AI tools that are not only technically robust but also practically impactful.

Research Interests

Computer Vision: Image classification, Semantic segmentation, Medical image analysis
Deep Learning: Self-supervised learning, Generative modeling, Explainable AI (XAI)

Publications

medRxiv

[Preprint] An Unsupervised XAI Framework for Dementia Detection with Context Enrichment

Devesh Singh, Yusuf Brima, Fedor Levin, Martin Becker, Bjarne Hiller, Andreas Hermann, Irene Villar-Munoz, Lukas Beichert, Alexander Bernhardt, Katharina Buerger, Michaela Butryn, Peter Dechent, Emrah Duezel, Michael Ewers, Klaus Fliessbach, Silka D. Freiesleben, Wenzel Glanz, Stefan Hetzer, Daniel Janowitz, Doreen Goerss, Ingo Kilimann, Okka Kimmich, Christoph Laske, Johannes Levin, Andrea Lohse, Falk Luesebrink, Matthias Munk, Robert Perneczky, Oliver Peters, Lukas Preis, Josef Priller, Johannes Prudlo, Diana Prychynenko, Boris-Stephan Rauchmann, Ayda Rostamzadeh, Nina Roy-Kluth, Klaus Scheffler, Anja Schneider, Louise Droste zu Senden, Bjoern H. Schott, Annika Spottke, Matthis Synofzik, Jens Wiltfang, Frank Jessen, Marc-Andre Weber, Stefan J Teipel, Martin Dyrba* (*Corresponding author)

medRxiv

DOI Code

CIBM

SMAS: Structural MRI-based AD Score using Bayesian supervised VAE

Aditya Nemali, Jose Bernal, Renat Yakupov, Devesh Singh, Martin Dyrba, EI Incesoy, Sach Mukherjee, Oliver Peters, J Hellmann-Regen, Ersin Ersoezlue, Lukas Preis, Josef Priller, Eike Spruth, Slawek Altenstein, Andrea Lohse, Anja Schneider, Klaus Fliessbach, Okka Kimmich, Jens Wiltfang, Niels Hansen, Bjoernhendrik Schott, Ayda Rostamzadeh, Wenzel Glanz, Michaela Butryn, Katharina Buerger, Daniel Janowitz, Michael Ewers, Robert Perneczky, Boris Rauchmann, Stefan Teipel, Ingo Kilimann, Doreen Goerss, Christoph Laske, Sebastian Sodenkamp, Annika Spottke, Marie Coenjaerts, Frederic Brosseron, Falk Luesebrink, Peter Dechent, Klaus Scheffler, Stefan Hetzer, Luca Kleineidam, Melina Stark, Frank Jessen, E Duzel, Gabriel Ziegler

Computers in Biology and Medicine

DOI

BVM

Evaluating the fidelity of explanations for convolutional neural networks in alzheimer’s disease detection

Bjarne C Hiller, Sebastian Bader, Devesh Singh, Thomas Kirste, Martin Becker, Martin Dyrba

German Conference on Medical Image Computing (BVM)

DOI Code

JAD

A computational ontology framework for the synthesis of multi-level pathology reports from brain MRI scans

Devesh Singh, Alice Grazia, Achim Reiz, Andreas Hermann, Slawek Altenstein, Lukas Beichert, Alexander Bernhardt, Katharina Buerger, Michaela Butryn, Peter Dechent, Emrah Duezel, Michael Ewers, Klaus Fliessbach, Silka D Freiesleben, Wenzel Glanz, Stefan Hetzer, Daniel Janowitz, Ingo Kilimann, Okka Kimmich, Christoph Laske, Johannes Levin, Andrea Lohse, Falk Luesebrink, Matthias Munk, Robert Perneczky, Oliver Peters, Lukas Preis, Josef Priller, Johannes Prudlo, Boris S Rauchmann, Ayda Rostamzadeh, Nina Roy-Kluth, Klaus Scheffler, Anja Schneider, Luisa S Schneider, Björn H Schott, Annika Spottke, Eike J Spruth, Matthis Synofzik, Jens Wiltfang, Frank Jessen, Stefan J Teipel, Martin Dyrba* (*Corresponding author)

Journal of Alzheimer’s Disease

DOI Code

Front. Neuroinform.

Contrastive self-supervised learning for neurodegenerative disorder classification

Vadym Gryshchuk, Devesh Singh, Stefan Teipel, Martin Dyrba* (*Corresponding author)

Frontiers in Neuroinformatics

DOI Code

BVM

Comparison of CNN Architectures for Detecting Alzheimer’s Disease using Relevance Maps

Devesh Singh, Martin Dyrba

German Conference on Medical Image Computing (BVM)

DOI Code

Academic Services

Contact Details

Address: Ernst-Heydemann-Str. 6, 18057 Rostock, Germany
Office Location: Universitätsmedizin Rostock: Biomedicum
Email: devesh.singh (at) uni-rostock.de