Photo of Milda Poceviciute

Milda Poceviciute


Research area: digital pathology, deep learning, XAI, anomaly detection.

Explainability and robustness in digital pathology 

Milda Pocevičiūtė is a Ph.D. candidate at Linköping University working in the Computer Graphics and Image processing group under supervision of Claes Lundström and Gabriel Eilertsen. The previous education includes MSc Statistics and Machine learning from the Linköping University (acquired in 2019).  
The research focus is on digital pathology and what explainable artificial intelligence (XAI), anomaly detection and uncertainty techniques are necessary for bringing state-of-the-art deep learning algorithms to the clinical practice. 



Milda Poceviciute, Gabriel Eilertsen, Claes Lundström (2024) Benefits of spatial uncertainty aggregation for segmentation in digital pathology Journal of Medical Imaging, Vol. 11 Continue to DOI


Milda Poceviciute, Gabriel Eilertsen, Claes Lundström (2023) Spatial uncertainty aggregation for false negatives detection in breast cancer metastases segmentation MEDICAL IMAGING 2023, Article 124710W Continue to DOI
Milda Pocevičiūtė (2023) Generalisation and reliability of deep learning for digital pathology in a clinical setting
Milda Pocevičiūtė, Gabriel Eilertsen, Stina Garvin, Claes Lundström (2023) Detecting Domain Shift in Multiple Instance Learning for Digital Pathology Using Fréchet Domain Distance Medical Image Computing and Computer Assisted Intervention – MICCAI 2023: 26th International Conference, Vancouver, BC, Canada, October 8–12, 2023, Proceedings, Part V, p. 157-167 Continue to DOI


Sofia Jarkman, Micael Karlberg, Milda Poceviciute, Anna Bodén, Peter Bandi, Geert Litjens, Claes Lundström, Darren Treanor, Jeroen van der Laak (2022) Generalization of Deep Learning in Digital Pathology: Experience in Breast Cancer Metastasis Detection Cancers, Vol. 14, Article 5424 Continue to DOI