Publications

Bias in Unsupervised Anomaly Detection in Brain MRI

UAD faces significant challenges tied to biases from different sources, including scanners, sex, and race. Future UAD models should maintain sensitivity to pathological shifts while minimizing sensitivity to non-pathological factors.

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

Bias in Unsupervised Anomaly Detection in Brain MRI

Mask, Stitch, and Re-Sample: Enhancing Robustness and Generalizability in Anomaly Detection through Automatic Diffusion Models

Overcoming the noise paradox of diffusion models

Cosmin I. Bercea, Michael Neumayr, Daniel Rueckert, Julia Schnabel

Generalizing Unsupervised Anomaly Detection: Towards Unbiased Pathology Screening

Moving beyond hyperintensity thresholding. This paper analyzes the challenges and outlines opportunities for advancing the field of unsupervised anomaly detection.

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

Generalizing Unsupervised Anomaly Detection: Towards Unbiased Pathology Screening

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

Generative networks to reverse anomalies in medical imaging.

Cosmin I. Bercea, Benedikt Wiestler, Daniel Rueckert, Julia Schnabel

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

Federated Disentangled Representation Learning for Unsupervised Brain Anomaly Detection

Implicit disentanglement of shape and appearance with federated learning for unsupervised brain pathology segmentation.

Cosmin I. Bercea, Benedikt Wiestler, Daniel Rückert, Shadi Albarqouni

Federated Disentangled Representation Learning for Unsupervised Brain Anomaly Detection

What do we learn? Debunking the Myth of Unsupervised Outlier Detection

Novel deformable auto-encoders for unsupervised outlier detection

Cosmin I. Bercea, Daniel Rueckert, Julia Schnabel

What do we learn? Debunking the Myth of Unsupervised Outlier Detection

FedDis: Disentangled Federated Learning for Unsupervised Brain Pathology Segmentation

Implicit disentanglement of shape and appearance with federated learning for unsupervised brain pathology segmentation.

Cosmin I. Bercea, Benedikt Wiestler, Daniel Rückert, Shadi Albarqouni

SHAMANN: Shared Memory Augmented Neural Networks

Multiple virtual actors cooperating through shared memory solve medical image segmentation.

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

SHAMANN: Shared Memory Augmented Neural Networks

Confidence-aware Levenberg-Marquardt optimization for joint motion estimation and super-resolution

Multi-scale robust super-resolution by jointly optimizing the motion estimation and image reconstruction.

Cosmin I. Bercea, Andreas K. Maier, Thomas Koehler

Confidence-aware Levenberg-Marquardt optimization for joint motion estimation and super-resolution