TY  - JOUR
AU  - Chatterjee, Soumick
AU  - Gaidzik, Franziska
AU  - Sciarra, Alessandro
AU  - Mattern, Hendrik
AU  - Janiga, Gábor
AU  - Speck, Oliver
AU  - Nürnberger, Andreas
AU  - Pathiraja, Sahani
TI  - PULASki: Learning inter-rater variability using statistical distances to improve probabilistic segmentation.
JO  - Medical image analysis
VL  - 103
SN  - 1361-8415
CY  - Amsterdam [u.a.]
PB  - Elsevier Science
M1  - DZNE-2025-00656
SP  - 103623
PY  - 2025
AB  - In the domain of medical imaging, many supervised learning based methods for segmentation face several challenges such as high variability in annotations from multiple experts, paucity of labelled data and class imbalanced datasets. These issues may result in segmentations that lack the requisite precision for clinical analysis and can be misleadingly overconfident without associated uncertainty quantification. This work proposes the PULASki method as a computationally efficient generative tool for biomedical image segmentation that accurately captures variability in expert annotations, even in small datasets. This approach makes use of an improved loss function based on statistical distances in a conditional variational autoencoder structure (Probabilistic UNet) , which improves learning of the conditional decoder compared to the standard cross-entropy particularly in class imbalanced problems. The proposed method was analysed for two structurally different segmentation tasks (intracranial vessel and multiple sclerosis (MS) lesion) and compare our results to four well-established baselines in terms of quantitative metrics and qualitative output. These experiments involve class-imbalanced datasets characterised by challenging features, including suboptimal signal-to-noise ratios and high ambiguity. Empirical results demonstrate the PULASKi method outperforms all baselines at the 5
KW  - Humans
KW  - Observer Variation
KW  - Magnetic Resonance Imaging: methods
KW  - Algorithms
KW  - Image Interpretation, Computer-Assisted: methods
KW  - Multiple Sclerosis: diagnostic imaging
KW  - Supervised Machine Learning
KW  - Conditional VAE (Other)
KW  - Distribution distance (Other)
KW  - Multiple sclerosis segmentation (Other)
KW  - Probabilistic UNet (Other)
KW  - Vessel segmentation (Other)
LB  - PUB:(DE-HGF)16
C6  - pmid:40367700
DO  - DOI:10.1016/j.media.2025.103623
UR  - https://pub.dzne.de/record/278930
ER  -