BERCEA

Curriculum Vitae

Postdoctoral Researcher @ TUM • AI & Medical Imaging

Download PDF ↓

Honors & Awards

2024

2nd Place, PhD Thesis Madness

MICCAI 2024, Marrakesh

"Global competition for groundbreaking doctoral research in medical imaging."

2024

Best Paper Award

EMERGE & ADSM Workshops @ MICCAI

"Recognized for work on Deformable Autoencoders and Test-Time Adaptation."

Academic Service

2024 - 2026

Scientific Event Officer

MICCAI Student Board

Venue Impact

Elected officer representing the global student community and coordinating scientific events.

2025

Session and Area Chair

MICCAI

Venue Impact

Coordinating peer-reviews and chairing oral sessions for the premier medical imaging conference.

2025

Program Chair

MedEurIPS @ NeurIPS 2025

Venue Impact

Leading the flagship Medical Imaging workshop at EurIPS to bridge ML and clinical research.

2025

Lead Organizer

MICCAI EMERGE Workshop

Venue Impact

Directing the MICCAI workshop focused on Foundation Models and Generative AI in Medical Imaging.

2025

Co-organizer

MICCAI MOOD Challenge

Venue Impact

Directing the Medical Out-of-Distribution analysis challenge to benchmark anomaly detection robustness.

Active

Reviewer

MICCAI, IEEE TMI, MEDIA

Venue Impact

Providing expert peer review for top-tier journals and conferences in AI and Medical Imaging.

Leadership & Teaching

Course Tutor

AI in Medicine

since 2023

Seminar Lead

AI for Vision and Multimodal Learning in Medical Imaging

since 2024 @ TUM

Seminar Lead

Unsupervised Anomaly Detection in Medicine

2022 - 2024 @ TUM

Thesis Supervisor

Master's Theses and Projects in AI

Ongoing @ TUM

Publications

Google Scholar ↗

SegMaST: Mamba-based Spatio-Temporal Modeling to Improve Longitudinal Disease Detection and Segmentation

Authors: Aswathi Varma, Jonas Weidner, Laurin Lux, Cosmin Bercea, Mark Mühlau, Jan Kirschke, Benedikt Wiestler, Daniel Rueckert

Preprint [Under Review] • 2026
Knowledge to Sight: Reasoning over Visual Attributes via Knowledge Decomposition for Abnormality Grounding

Authors: Jun Li, Che Liu, Wenjia Bai, Mingxuan Liu, Rossella Arcucci, Cosmin I. Bercea, Julia A. Schnabel

WACV • 2026
NOVA: A Benchmark for Rare Anomaly Localization and Clinical Reasoning in Brain MRI

Authors: Cosmin I. Bercea, Jun Li, Philipp Raffler, Evamaria Olga Riedel, Lena Schmitzer, Angela Kurz, Felix Bitzer, Paula Roßmüller, Julian Canisius, Mirjam L Beyrle, others

NeurIPS Datasets and Benchmarks Track • 2025
Evaluating normative representation learning in generative AI for robust anomaly detection in brain imaging

Authors: Cosmin I. Bercea, Benedikt Wiestler, Daniel Rückert, Julia A. Schnabel

Nature Communications • 2025
Denoising diffusion models for anomaly localization in medical images

Authors: Cosmin I. Bercea, Philippe C. Cattin, Julia A. Schnabel, Julia Wolleb

MELBA • 2025
Does DINOv3 set a new medical vision standard?

Authors: Che Liu, Yinda Chen, Haoyuan Shi, Jinpeng Lu, Bailiang Jian, Jiazhen Pan, Linghan Cai, Jiayi Wang, Yundi Zhang, Jun Li, Cosmin I. Bercea, Cheng Ouyang, Chen Chen, Zhiwei Xiong, Benedikt Wiestler, Christian Wachinger, Daniel Rueckert, Wenjia Bai, Rossella Arcucci

arXiv preprint • 2025
Enhancing Abnormality Grounding for Vision Language Models with Knowledge Descriptions

Authors: Jun Li, Che Liu, Wenjia Bai, Rossella Arcucci, Cosmin I. Bercea, Julia A. Schnabel

arXiv preprint • 2025
Semantic Alignment of Unimodal Medical Text and Vision Representations

Authors: Maxime Di Folco, Emily Chan, Marta Hasny, Cosmin I. Bercea, Julia A. Schnabel

arXiv preprint • 2025
Influence of Classification Task and Distribution Shift Type on OOD Detection in Fetal Ultrasound

Authors: Chun Kit Wong, Anders N Christensen, Cosmin I. Bercea, Julia A. Schnabel, Martin G Tolsgaard, Aasa Feragen

MICCAI • 2025
Learning to reason about rare diseases through retrieval-augmented agents

Authors: Ha Young Kim, Jun Li, Ana Beatriz Solana, Carolin M Pirkl, Benedikt Wiestler, Julia A. Schnabel, Cosmin I. Bercea

arXiv preprint • 2025
TomoGraphView: 3D Medical Image Classification with Omnidirectional Slice Representations and Graph Neural Networks

Authors: Johannes Kiechle, Stefan M Fischer, Daniel M Lang, Cosmin I. Bercea, Matthew J Nyflot, Lina Felsner, Julia A. Schnabel, Jan C Peeken

arXiv preprint • 2025
GroundingDINO for Open-Set Lesion Detection in Medical Imaging

Authors: Samuel James Roughley, Johanna Paula Müller, Shangqi Gao, Zeyu Gao, Marta Ligero, Rudolfs Blums, Mireia Crispin-Ortuzar, Julia A. Schnabel, Bernhard Kainz, Cosmin I. Bercea, Ines Machado

MICCAI EMERGE Workshop • 2025
Selective Test-Time Adaptation for Unsupervised Anomaly Detection using Neural Implicit Representations

Authors: Sameer Ambekar, Julia A. Schnabel, Cosmin I. Bercea

MICCAI ADSMI Workshop • 2024
Fmbench: Benchmarking fairness in multimodal large language models on medical tasks

Authors: Peiran Wu, Che Liu, Canyu Chen, Jun Li, Cosmin I. Bercea, Rossella Arcucci

arXiv preprint • 2024
Towards universal unsupervised anomaly detection in medical imaging

Authors: Cosmin I. Bercea, Benedikt Wiestler, Daniel Rueckert, Julia A. Schnabel

arXiv preprint • 2024
Diffusion models with implicit guidance for medical anomaly detection

Authors: Cosmin I. Bercea, Benedikt Wiestler, Daniel Rueckert, Julia A. Schnabel

MICCAI • 2024
Interpretable Representation Learning of Cardiac MRI via Attribute Regularization

Authors: Maxime Di Folco, Cosmin I. Bercea, Emily Chan, Julia A. Schnabel

MICCAI • 2024
Language models meet anomaly detection for better interpretability and generalizability

Authors: Jun Li, Su Hwan Kim, Philip Müller, Lina Felsner, Daniel Rueckert, Benedikt Wiestler, Julia A. Schnabel, Cosmin I. Bercea

MICCAI MMMI Workshop • 2024
Multi-image visual question answering for unsupervised anomaly detection

Authors: Jun Li, Cosmin I. Bercea, Philip Müller, Lina Felsner, Su Hwan Kim, Daniel Rueckert, Benedikt Wiestler, Julia A. Schnabel

arXiv preprint • 2024
Unsupervised Analysis of Alzheimer’s Disease Signatures using 3D Deformable Autoencoders

Authors: Mehmet Yigit Avci, Emily Chan, Veronika Zimmer, Daniel Rueckert, Benedikt Wiestler, Julia A. Schnabel, Cosmin I. Bercea

MICCAI EMERGE Workshop • 2024
Mededit: Counterfactual diffusion-based image editing on brain mri

Authors: Malek Ben Alaya, Daniel M Lang, Benedikt Wiestler, Julia A. Schnabel, Cosmin I. Bercea

MICCAI Sashimi Workshop • 2024
Diffusion models for unsupervised anomaly detection in fetal brain ultrasound

Authors: Hanna Mykula, Lisa Gasser, Silvia Lobmaier, Julia A. Schnabel, Veronika Zimmer, Cosmin I. Bercea

MICCAI ASMUS Workshop • 2024
Denoising Diffusion Models for 3D Healthy Brain Tissue Inpainting

Authors: Alicia Durrer, Julia Wolleb, Florentin Bieder, Paul Friedrich, Lester Melie-Garcia, Mario Alberto Ocampo Pineda, Cosmin I. Bercea, Ibrahim Ethem Hamamci, Benedikt Wiestler, Marie Piraud, Oezguer Yaldizli, Cristina Granziera, Bjoern Menze, Philippe C Cattin, others

MICCAI DGM4 Workshop • 2024
Attribute Regularized Soft Introspective VAE: Towards Cardiac Attribute Regularization Through MRI Domains

Authors: Maxime Di Folco, Cosmin Bercea, Julia A. Schnabel

arXiv preprint • 2023
Mask, stitch, and re-sample: Enhancing robustness and generalizability in anomaly detection through automatic diffusion models

Authors: Cosmin I. Bercea, Michael Neumayr, Daniel Rueckert, Julia A. Schnabel

ICML IMLH Workshop • 2023
Generalizing Unsupervised Anomaly Detection: Towards Unbiased Pathology Screening

Authors: Cosmin I. Bercea, Benedikt Wiestler, Daniel Rueckert, Julia A. Schnabel

Medical Imaging with Deep Learning • 2023
What do AEs learn? Challenging common assumptions in unsupervised anomaly detection

Authors: Cosmin I. Bercea, Daniel Rueckert, Julia A. Schnabel

MICCAI • 2023
Reversing the Abnormal: Pseudo-Healthy Generative Networks for Anomaly Detection

Authors: Cosmin I. Bercea, Benedikt Wiestler, Daniel Rueckert, Julia A. Schnabel

MICCAI • 2023
Bias in unsupervised anomaly detection in brain MRI

Authors: Cosmin I. Bercea, Esther Puyol-Antón, Benedikt Wiestler, Daniel Rueckert, Julia A. Schnabel, Andrew P King

MICCAI FAIMI Workshop • 2023
Multispectral 3D masked autoencoders for anomaly detection in non-contrast enhanced breast MRI

Authors: Daniel M Lang, Eli Schwartz, Cosmin I. Bercea, Raja Giryes, Julia A. Schnabel

MICCAI Workshop on Cancer Prevention through Early Detection • 2023
Federated disentangled representation learning for unsupervised brain anomaly detection

Authors: Cosmin I. Bercea, Benedikt Wiestler, Daniel Rueckert, Shadi Albarqouni

Nature Machine Intelligence • 2022
SPA: Shape-Prior Variational Autoencoders for Unsupervised Brain Pathology Segmentation

Authors: Cosmin I. Bercea, Benedikt Wiestler, Daniel Rueckert, Shadi Albarqouni

Preprint • 2022
Feddis: Disentangled federated learning for unsupervised brain pathology segmentation

Authors: Cosmin I. Bercea, Benedikt Wiestler, Daniel Rueckert, Shadi Albarqouni

arXiv preprint • 2021
SHAMANN: Shared memory augmented neural networks

Authors: Cosmin I. Bercea, Olivier Pauly, Andreas Maier, Florin C Ghesu

IPMI • 2019
Confidence-aware levenberg-marquardt optimization for joint motion estimation and super-resolution

Authors: Cosmin Bercea, Andreas Maier, Thomas Köhler

IEEE ICIP • 2016

Patents

US Patent 12,300,030

2025

Detection device for recognizing an object and/or a person

Bercea, C.I., Kapelner, T.

US Patent 12,243,241

2025

Tracking of multiple objects using local and shared memory

Bercea, C.I.

US Patent 12,086,993

2024

Tracking in cooperation with multiple neural networks

Bercea, C.I.

US Patent 10,885,663

2021

Method for setting a viewing direction in a virtual environment

Mutschler, C., et al.

US Patent 10,825,172

2020

Medical image segmentation

Pauly, O., Ghesu, F.C., Bercea, C.I.

DE Patent DE102020212856A1

2020

Method and device for modeling a context of a physical system

Lengsfeld, A., et al.

Experience

2024 - present
Postdoctoral Researcher

Leading research on Generative AI and Multimodal Learning for medical image analysis to identify rare and early-stage diseases.

Technical University of Munich (TUM)

Munich, Germany
2020 - 2024
PhD Researcher

Dissertation: "Deep Generative Models for Unsupervised Anomaly Detection in Medical Imaging." Developed SOTA algorithms for pathology screening.

Technical University of Munich (TUM)

Munich, Germany
2018 - 2020
Research Engineer

R&D of deep learning models for perception and behavior prediction in autonomous driving systems.

Bosch Corporate Research

Hildesheim, Germany
2017 - 2018
Research Assistant

Focused on Machine Learning for Medical Imaging under the supervision of Dr. Florin C. Ghesu. and Dr. Oliver Pauly.

Siemens Healthineers

Erlangen, Germany
2013 - 2017
Research Assistant

Research in Machine Learning and Virtual Reality supervised by Dr.-Ing. Christopher Mutschler.

Fraunhofer IIS

Erlangen, Germany
2016
Research Assistant

Research on Machine Learningf for Action Recognition supervised by Prof.Dr. Petia Radeva.

Computer Vision Center (UAB)

Barcelona, Spain

Education

2020 - 2024
PhD (Dr. rer. nat.) in Computer Science

Focus on Deep Generative Models. Supervised by Prof. Julia Schnabel and Prof. Daniel Rückert.

Technical University of Munich (TUM)

Munich, Germany
2023
Visiting Researcher

Research on Biases in unsupervised learning with Dr. Andrew King and Dr. Esther Puyol-Antón.

King's College London

London, UK
2011 - 2018
B.Sc & M.Sc. in Computer Science

Grade 1.6. Specialization in Pattern Recognition and Medical Image Analysis under Prof. Dr. Andreas Maier.

FAU Erlangen-Nuremberg

Erlangen, Germany

Honors & Awards

2nd Place, PhD Thesis Madness

Best Paper Award

Academic Service

Scientific Event Officer

Session and Area Chair

Program Chair

Lead Organizer

Co-organizer

Reviewer

Leadership & Teaching

AI in Medicine

AI for Vision and Multimodal Learning in Medical Imaging

Unsupervised Anomaly Detection in Medicine

Master's Theses and Projects in AI

Publications

SegMaST: Mamba-based Spatio-Temporal Modeling to Improve Longitudinal Disease Detection and Segmentation

Knowledge to Sight: Reasoning over Visual Attributes via Knowledge Decomposition for Abnormality Grounding

NOVA: A Benchmark for Rare Anomaly Localization and Clinical Reasoning in Brain MRI

Evaluating normative representation learning in generative AI for robust anomaly detection in brain imaging

Denoising diffusion models for anomaly localization in medical images

Does DINOv3 set a new medical vision standard?

Enhancing Abnormality Grounding for Vision Language Models with Knowledge Descriptions

Semantic Alignment of Unimodal Medical Text and Vision Representations

Influence of Classification Task and Distribution Shift Type on OOD Detection in Fetal Ultrasound

Learning to reason about rare diseases through retrieval-augmented agents

TomoGraphView: 3D Medical Image Classification with Omnidirectional Slice Representations and Graph Neural Networks

GroundingDINO for Open-Set Lesion Detection in Medical Imaging

Selective Test-Time Adaptation for Unsupervised Anomaly Detection using Neural Implicit Representations

Fmbench: Benchmarking fairness in multimodal large language models on medical tasks

Towards universal unsupervised anomaly detection in medical imaging

Diffusion models with implicit guidance for medical anomaly detection

Interpretable Representation Learning of Cardiac MRI via Attribute Regularization

Language models meet anomaly detection for better interpretability and generalizability

Multi-image visual question answering for unsupervised anomaly detection

Unsupervised Analysis of Alzheimer’s Disease Signatures using 3D Deformable Autoencoders

Mededit: Counterfactual diffusion-based image editing on brain mri

Diffusion models for unsupervised anomaly detection in fetal brain ultrasound

Denoising Diffusion Models for 3D Healthy Brain Tissue Inpainting

Attribute Regularized Soft Introspective VAE: Towards Cardiac Attribute Regularization Through MRI Domains

Mask, stitch, and re-sample: Enhancing robustness and generalizability in anomaly detection through automatic diffusion models

Generalizing Unsupervised Anomaly Detection: Towards Unbiased Pathology Screening

What do AEs learn? Challenging common assumptions in unsupervised anomaly detection

Reversing the Abnormal: Pseudo-Healthy Generative Networks for Anomaly Detection

Bias in unsupervised anomaly detection in brain MRI

Multispectral 3D masked autoencoders for anomaly detection in non-contrast enhanced breast MRI

Federated disentangled representation learning for unsupervised brain anomaly detection

SPA: Shape-Prior Variational Autoencoders for Unsupervised Brain Pathology Segmentation

Feddis: Disentangled federated learning for unsupervised brain pathology segmentation

SHAMANN: Shared memory augmented neural networks

Confidence-aware levenberg-marquardt optimization for joint motion estimation and super-resolution

Patents

Detection device for recognizing an object and/or a person

Tracking of multiple objects using local and shared memory

Tracking in cooperation with multiple neural networks

Method for setting a viewing direction in a virtual environment

Medical image segmentation

Method and device for modeling a context of a physical system

Experience

Education